Prioritisation of identified ChatGPT enablers towards developing Sustainable Healthcare Systems: A Fuzzy MCDM approach

Abstract

Recent advances in Artificial Intelligence (AI) applications to healthcare, especially through intelligent conversational tools like ChatGPT, have opened new paths to support sustainability, operational effectiveness, and patient-centred care. However, the successful implementation of these technologies very much relies on the proper identification and prioritisation of these enablers, as they are responsible for the successful implementation in healthcare systems. This study aims to prioritise identified ChatGPT enablers that lead to sustainable healthcare systems. In this study hybrid Fuzzy Multi-Criteria Decision-Making (FMCDM) approach suitably employed that integrates fuzzy Analytic hierarchy process (FAHP), fuzzy Multiple Criteria Ranking by Alternative Trace (FMCRAT) and fuzzy Ranking of Alternatives by Perimeter Similarity (FRAPS). Expert opinions were collected using a structured questionnaire, and fuzzy logic was applied to manage uncertainty and imprecision in human judgments. The results revealed that Patient Feedback, Clinical Decision Support, and Medical Queries are the most significant enablers facilitating sustainable adoption of ChatGPT in healthcare. However, Telemedicine and Fraud Detection were found to have lower influence, suggesting that although their contributions are limited, they serve as enabling factors that support the overall healthcare framework. Finally, a sensitivity analysis was conducted to assess the robustness of the findings, confirming the consistency and stability of the proposed framework. The research outcomes contribute both theoretically and practically. Theoretically, it extends the application of FMCDM methods to AI-driven healthcare decision-making. Practically, it provides a structured decision-support model to help policymakers, healthcare administrators, and developers for optimising resources and developing evidence-based initiatives for ChatGPT adoption. The study concludes that strengthening top-ranked enablers can significantly enhance healthcare sustainability, patient engagement, and system performance through responsible AI adoption. This article provides a methodological contribution by a novel approach through integration of various fuzzy ranking methods to cross-verify the results of prioritisation and increase the methodological strength of AI-driven medical decision support systems.

Graphical abstract

Keywords

ChatGPT / Sustainable healthcare / AI technologies / fuzzy MCDM / fuzzy AHP / fuzzy MCRAT / fuzzy RAPS / Sensitivity Analysis

1 Introduction

The rapid adoption of Artificial Intelligence (AI) technologies has significantly transformed healthcare systems by enhancing efficiency, accuracy, and accessibility of services. Among these technologies, conversational AI tools such as ChatGPT have attracted growing attention due to their potential to support intelligent communication, clinical decision support, administrative automation, and patient engagement. By assisting healthcare professionals with tasks such as clinical documentation, data retrieval, and patient education, ChatGPT can improve operational performance and contribute to the long-term sustainability of healthcare systems [1,2].

Sustainable healthcare emphasises the efficient utilisation of resources, minimisation of waste, and consistent delivery of high-quality care over time. Increasing operational costs, workforce shortages, and rising patient demands have placed considerable pressure on healthcare systems. In this context, AI-enabled solutions such as ChatGPT offer opportunities to enhance system resilience, operational efficiency, and innovation [3]. However, the sustainable integration of ChatGPT in healthcare is influenced by multiple interrelated factors, including data privacy, system reliability, cost-effectiveness, accuracy, and organisational collaboration. A lack of careful evaluation of these factors may limit the long-term benefits of ChatGPT deployment in healthcare facilities [4].

Although prior studies have extensively examined the adoption of AI technologies in healthcare, limited research has explicitly focused on ChatGPT as a facilitator of sustainable healthcare systems. Existing studies primarily address AI implementation from technological, clinical, or ethical perspectives, without offering an integrated evaluation framework that aligns sustainability objectives with decision-support methodologies [5,6]. Moreover, most existing evaluations rely on conventional decision-making techniques that assume precise judgments, which may not adequately capture the ambiguity and subjectivity inherent in expert assessments within complex healthcare systems [7].

This highlights a critical research gap in the application of Fuzzy Multi-Criteria Decision-Making (FMCDM) methods to systematically evaluate and prioritise ChatGPT enablers across technical, organisational, and sustainability dimensions. Fuzzy-based approaches are well suited to healthcare decision-making, as they allow for the incorporation of linguistic assessments and uncertainty in expert opinions [8,9]. Despite their suitability, FMCDM techniques have not been sufficiently employed to identify the most influential enablers for the sustainable implementation of ChatGPT in healthcare systems.

To address this gap, the present study proposes a structured decision-support framework based on Fuzzy Multiple Criteria Ranking by Alternative Trace (FMCRAT) and Fuzzy Ranking of Alternatives by Perimeter Similarity (FRAPS). These methods enable robust prioritisation of ChatGPT enablers while effectively managing uncertainty and subjectivity in expert evaluations. Accordingly, the following research questions are formulated:

RQ 1: What are the key enablers of ChatGPT toward the adoption for developing sustainable healthcare systems?

RQ 2: How can the identified enablers be systematically evaluated and prioritised to support effective and sustainable implementation of ChatGPT in healthcare?

RQ 3: How does the application of a Fuzzy MCRAT and Fuzzy RAPS approach help manage uncertainty and subjectivity in prioritising ChatGPT enablers?

RQ 4: Which enablers are most critical for ensuring the long-term sustainability, efficiency, and reliability of ChatGPT integration in healthcare systems?

This study contributes to the literature by providing a comprehensive and systematic framework for identifying and prioritising ChatGPT enablers within sustainable healthcare systems. By integrating expert knowledge with FMCRAT and FRAPS methodologies, the proposed approach enhances decision-making accuracy under uncertainty. The findings offer practical insights for healthcare policymakers and administrators to allocate resources efficiently, formulate informed strategies, and support the ethical and sustainable implementation of ChatGPT to improve healthcare system performance and patient interactions.

The remainder of this paper is organised as follows. Section 2 reviews the relevant literature on ChatGPT enablers and evaluation criteria for sustainable healthcare system. Section 3 presents the research methodology, including expert involvement and the application of FAHP, FMCRAT, and FRAPS methods. Section 4 describes the data collection process, results, and sensitivity analysis. Sections 5 and 6 discuss the findings and their practical implications, respectively. Section 7 outlines the study’s limitations and future research directions, and Section 8 concludes the paper with key insights and recommendations.

2 Identification of Chat GPT Enablers and evaluation criteria for Sustainable Healthcare Systems from Literature

An extensive literature search across databases, including Scopus, Web of Science, Google Scholar, and PubMed, was conducted to identify potential ChatGPT for sustainable healthcare systems. Keywords that were used in search included “sustainable healthcare”, “ChatGPT enablers”, “enablers”, “telemedicine technology”, “intelligent healthcare”, “AI enablers”, and “Medical 4.0”. Retrieved data were initially screened based on titles and abstracts to remove duplicates and studies not relevant to AI-enabled healthcare or sustainability considerations. The full texts of the remaining studies were assessed for eligibility by examining their explicit discussion of ChatGPT or related AI technologies and their relevance to healthcare sustainability. From the eligible studies, enablers and evaluation criteria were systematically extracted, synthesised, and consolidated by merging conceptually similar factors. The resulting list was subsequently validated and refined through expert consultation to ensure clarity and contextual relevance. This process led to the final selection of nine ChatGPT enablers and nine evaluation criteria for further analysis. Figure 1 presents the conceptual framework of the study, illustrating the finalised ChatGPT enablers and the evaluation criteria used to assess their contribution to sustainable healthcare systems.

2.1 ChatGPT Enablers for Sustainable Healthcare Systems

Strategic planning is one of the primary outcomes of prioritisation, providing healthcare leaders with a coordinated roadmap for implementing ChatGPT enablers in leading sustainable healthcare systems. It ensures that implementation efforts remain aligned with the organisation’s long-term goals and overall vision. Determining the appropriate number of components is crucial for effective project decisions, as it helps manage complexity, optimise resource use, enhance clarity in decision-making, improve communication, and reduce potential biases [10,11]. As described in Table 1, the finalised set of ChatGPT enablers reflects these considerations for guiding sustainable healthcare implementation.

2.2 Criteria considered for evaluation of Chat GPT enablers

The identification of criteria encompasses all vital dimensions of decision-making challenges within the context of sustainable healthcare systems, ensuring that no critical factors are neglected. These criteria ensure a balanced assessment of how effectively each enabler contributes to system performance, patient engagement, data governance, and long-term sustainability. For analytical clarity, the criteria were categorized into beneficial and non-beneficial types. Beneficial criteria represent attributes where higher values indicate improved performance, such as reliability (C1), data privacy (C3), sustainability (C4), problem solving (C6), accuracy (C7), availability (C8), and collaboration (C9). In contrast, non-beneficial criteria correspond to attributes where lower values are preferred, including risk (C2), and cost-effectiveness (C5). This categorisation supports a transparent and robust application of MCDM techniques for prioritising ChatGPT enablers in sustainable healthcare systems. Table 2 provides a detailed description of these evaluation criteria.

3 Research Methodology

The study undertakes an in-depth literature review to develop a comprehensive list of potential enablers for ChatGPT implementation in healthcare, along with an outline of appropriate evaluation criteria. Based on best practices in technology evaluation research, an expert consultation process was conducted to narrow and authenticate the identified enablers and evaluation criteria from the literature. The prioritisation of ChatGPT enablers involves inherent complexity due to the multifaceted interactions among evaluation criteria and the need to accommodate diverse performance dimensions simultaneously, which traditional unidimensional analyses cannot adequately address [6,61].

The FMCDM techniques are well explored in the literature for responding to uncertainty in the judgment by experts and providing influential decision support in scenarios described by limited sample sizes and in linguistically expressed measurements [62,63]. Specifically, this study proposes an integrated application of the FAHP, FMCRAT, and FRAPS to obtain criterion weights and systematically rank ChatGPT enablers.

In the first stage, FAHP was used to stimulate relative criterion weights based on pairwise comparisons under a fuzzy linguistic scale, allowing experts to express subjective judgments with inherent ambiguity [64]. In the subsequent stage, FMCRAT and FRAPS were employed to assess and prioritise the ChatGPT enablers, using the criterion weights derived through FAHP, and incorporating Trace and Perimeter Similarity measures to enhance discriminant validity. The integrated methodology aligns with established MCDM protocols that emphasise hierarchical weighting, followed by alternative ranking [65].

In line with existing methods for conducting expert-based MCDM studies, the analysis was reinforced by five qualified and experienced domain experts, whose input provided the empirical foundation for the model estimation [66]. To simplify data collection, a questionnaire was drafted to meet the methodological requirements of FAHP, FMCRAT, and FRAPS, including a fuzzy linguistic scale to reflect levels of expert opinion [67]. The questionnaire was administered via Google Forms, with the survey link shared with the group via email to collect responses. These expert responses were aggregated in order to develop the initial decision matrix, which was later incorporated into an analytical model. Prioritisation results were computed using Trace and Perimeter Similarity measures, and sensitivity analysis (SA) were conducted to ensure the reliability of the findings. Figure 2 represents the detailed methodological framework adopted in this study, and the professional profiles of the participating experts are summarised in Table 3. This study concludes with a comprehensive discussion of the findings, theoretical and practical implications, limitations, and potential pathways for future research.

3.1 Fuzzy AHP

Analytic hierarchy process (AHP) has been widely used as a practical MCDM tool or a weight estimation technique in many areas such as selection, evaluation, planning and development, decision making, forecasting [68,69]. The traditional AHP requires crisp judgments. However, due to the complexity and uncertainty involved in real-world decision problems, a decision maker may sometimes feel more confident providing fuzzy judgments than crisp comparisons [70]. The steps of the FAHP for the calculation of the weight of criteria are presented as follows [71]:

Step 1: In the first step, create a fuzzy-structured pairwise comparison matrix as given in equation (1).

\begin{array}{}\text{(1)}& \tilde{A}\text{=}\left[{\tilde{a}}_{11},{\tilde{a}}_{12}\dots {\tilde{a}}_{1n},{\tilde{a}}_{21},{\tilde{a}}_{22},\dots {\tilde{a}}_{2n};{\tilde{a}}_{n1},{\tilde{a}}_{n2}\dots {\tilde{a}}_{n3}\right]\end{array}

where {\tilde{\text{a}}}_{\text{xy}}=\left({\text{a}}_{\text{xy}}^{\text{l}},{\text{ a}}_{\text{xy}}^{\text{m}},{\text{ a}}_{\text{xy}}^{\text{u}}\right), indicates the importance of criteria ‘x’ in comparison to criteria ‘y’ in terms of triangular fuzzy number (TFN). The variables x and y have values ranging from 1 to n.

The expert would provide the relative importance of the criteria on a linguistic scale. The Linguistic scale is converted into TFN by using Table 4 [72].

Step 2: Compute the geometric mean by using equation 2.

\begin{array}{}\text{(2)}& \begin{array}{c}{\tilde{g}}_x\text{=}\left((g_x^l,g_x^m,g_x^u\right)\text{=}\\ \left({(b_{x1}^l\times b_{x2}^l\dots \dots \times b_{xn}^l)}^{\frac{1}{n}},{\left(b_{x1}^m\times b_{x2}^m\dots \dots \times b_{xn}^m\right)}^{\frac{1}{n}},{(b_{x1}^u\times b_{x2}^u\dots \times b_{xn}^u)}^{\frac{1}{n}}\right)\end{array}\end{array}

Step 3: Compute the fuzzy weight by using equation 3.

\begin{array}{}\text{(3)}& \begin{array}{c}{\tilde{f}}_x=\left(f_x^l,f_x^m,f_x^u\right)=\\ \left(\frac{g_x^l}{g_1^u+g_2^u+\dots +g_n^u},\frac{g_x^m}{g_1^m+g_2^m+\dots .+g_n^m}, \frac{g_x^u}{g_1^l+g_2^l+\dots +g_n^l}\right)\end{array}\end{array}

Step 4: Compute the consistency ratio by using equation 4, which is the defuzzified value of the fuzzy weight.

\begin{array}{}\text{(4)}& f_x^{*}= \frac{f_x^l+4f_x^m+f_x^u}{6}\end{array}

Now, the FMCRAT and FRAPS methods are employed to evaluate the relative importance of the enablers, as described in section 3.2.

3.2 FMCRAT and FRAPS

Multiple Criteria Ranking by Alternative Trace (MCRAT) and Ranking of Alternatives by Perimeter Similarity (RAPS), developed by Urošević et al., are novel MCDM techniques that employ matrix trace and perimeter similarity, respectively, to rank alternatives [73]. These methods offer several advantages, including conceptual clarity, rational decision logic, methodological versatility, and result reliability. Despite these strengths, only a limited number of studies have applied MCRAT and RAPS, and their use remains largely confined to material selection [74,75] and mining-related decision problems [73]. Moreover, as these techniques are relatively recent, existing studies have not adequately addressed analytical uncertainty inherent in expert-based evaluations. Since the conventional MCRAT method is unable to handle vagueness and imprecision in decision-makers’ judgments, its applicability is limited in complex and uncertain environments. To overcome this limitation and extend the applicability of MCRAT beyond mining and material selection contexts, this study develops fuzzy extensions of both the MCRAT and RAPS methods. Compared with other established FMCDM approaches, such as fuzzy MARCOS, fuzzy TOPSIS, fuzzy PROMETHEE, and fuzzy ELECTRE, the proposed FMCRAT and FRAPS methods involve fewer computational steps and offer a more straightforward evaluation process, while still producing reliable, rational, and robust results [73]. Consequently, the integration of fuzziness into MCRAT and RAPS enhances their ability to model real-world uncertainty and broadens their applicability to emerging decision-making domains, including sustainable healthcare systems.

The steps of the FMCRAT and FRAPS methods are presented as follows [76]:

Step 1: First, create an initial decision matrix. Using Linguistic data from Table 5, decision-makers evaluate each alternative’s performance under each criterion. The linguistic data are converted into Triangular fuzzy numbers (TFNs) using Table 5, as given below [77].

After converting the linguistic scale into a triangular fuzzy number (TFN) given by each decision maker, the aggregated fuzzy decision matrix is then created by integrating the fuzzy values provided by each decision maker using the arithmetic mean approach. The aggregated fuzzy decision matrix is shown in equation (5).

\begin{array}{}\text{(5)}& \tilde{\text{D}}\text{=}[{\tilde{\text{d}}}_{\text{ij}}{]}_{\text{n}\times \text{n}}\end{array}

Step 2: Equations (6) and (7) (for benefit and non-benefit criteria, respectively) are used to normalised the values in the aggregated fuzzy decision matrix.

\begin{array}{}\text{(6)}& {\tilde{\text{e}}}_{\text{ij}}=({\text{e}}_{\text{ij}}^{\text{l}},{\text{ e}}_{\text{ij}}^{\text{m}},{\text{ e}}_{\text{ij}}^{\text{u}})=\frac{{\tilde{\text{d}}}_{\text{ij}}}{{\text{max}}_{\text{i}}{\text{d}}_{\text{ij}}^{\text{u}}}=\left(\frac{{\text{d}}_{\text{ij}}^{\text{l}}}{{\text{max}}_{\text{i}}{\text{d}}_{\text{ij}}^{\text{u}}},\frac{{\text{d}}_{\text{ij}}^{\text{m}}}{{\text{max}}_{\text{i}}{\text{d}}_{\text{ij}}^{\text{u}}},\frac{{\text{d}}_{\text{ij}}^{\text{u}}}{{\text{max}}_{\text{i}}{\text{d}}_{\text{ij}}^{\text{u}}}\right)\end{array}

\begin{array}{}\text{(7)}& {\tilde{\text{e}}}_{\text{ij}}=\left({\text{e}}_{\text{ij}}^{\text{l}},{\text{ e}}_{\text{ij}}^{\text{m}},{\text{ e}}_{\text{ij}}^{\text{u}}\right)=\frac{{\text{min}}_{\text{i}}{\text{d}}_{\text{ij}}^{\text{l}}}{{\tilde{\text{d}}}_{\text{ij}}}=\left(\frac{{\text{min}}_{\text{i}}{\text{d}}_{\text{ij}}^{\text{l}}}{{\text{d}}_{\text{ij}}^{\text{u}}},\frac{{\text{min}}_{\text{i}}{\text{d}}_{\text{ij}}^{\text{m}}}{{\text{d}}_{\text{ij}}^{\text{m}}},\frac{{\text{min}}_{\text{i}}{\text{d}}_{\text{ij}}^{\text{u}}}{{\text{d}}_{\text{ij}}^{\text{l}}}\right)\end{array}

Step 3: The fuzzy normalised matrix is multiplied by the fuzzy weights of the criterion to get fuzzy weighted normalised matrix using equation (8).

\begin{array}{}\text{(8)}& {\tilde{\text{e}}}_{\text{ij}}=\left({\text{g}}_{\text{ij}}^{\text{l}},{\text{ g}}_{\text{ij}}^{\text{m}},{\text{ g}}_{\text{ij}}^{\text{u}}\right)={\text{f}}_{\text{x}}^{\text{*}}\times {\tilde{\text{e}}}_{\text{ij}}=\left({\text{f}}_{\text{x}}^{\text{*}}\times {\text{e}}_{\text{ij}}^{\text{l}},{\text{ f}}_{\text{x}}^{\text{*}}\times {\text{e}}_{\text{ij}}^{\text{m}},{\text{ f}}_{\text{x}}^{\text{*}}\times {\text{e}}_{\text{ij}}^{\text{u}}\right)\end{array}

Step 4: The fuzzy optimal alternative is determined using equations (9) and (10).

\begin{array}{}\text{(9)}& {\tilde{\text{p}}}_{\text{j}}=\text{max}({\tilde{\text{g}}}_{\text{ij}}|1\le \text{j}\le \text{n})\end{array}

\begin{array}{}\text{(10)}& \tilde{\text{p}}=\left\{{\tilde{\text{p}}}_1,{\tilde{\text{p}}}_2,\dots \dots \dots {\tilde{\text{p}}}_{\text{n}}\right\}\end{array}

Step 5: The fuzzy optimal alternatives are decomposed using equations (11) and (12).

\begin{array}{}\text{(11)}& \tilde{\text{P}}={\tilde{\text{P}}}^{\text{max}}\cup {\tilde{\text{P}}}^{\text{min}}\end{array}

\begin{array}{}\text{(12)}& \tilde{\text{P}}=\left\{{\tilde{\text{p}}}_1,{\tilde{\text{p}}}_2\dots \dots \dots {\tilde{\text{p}}}_{\text{k}}\right\}\cup \left\{{\tilde{\text{p}}}_1,{\tilde{\text{p}}}_2,\dots \dots \dots {\tilde{\text{p}}}_{\text{h}}\right\},\text{ k}+\text{h}=\text{j}\end{array}

Step 6: Alternatives are decomposed using equations (13) and (14).

\begin{array}{}\text{(13)}& \tilde{\text{R}}={\tilde{\text{R}}}^{\text{max}}\cup {\tilde{\text{R}}}^{\text{min}}\end{array}

\begin{array}{}\text{(14)}& \tilde{\text{R}}=\left\{{\tilde{\text{r}}}_1,{\tilde{\text{r}}}_2,\dots \dots \dots {\tilde{\text{r}}}_{\text{k}}\right\}\cup \left\{{\tilde{\text{r}}}_1,{\tilde{\text{r}}}_2,\dots \dots \dots {\tilde{\text{r}}}_{\text{h}}\right\},\text{ k}+\text{h}=\text{j}\end{array}

Step 7: The fuzzy magnitudes of the components are computed using equations (15–18).

\begin{array}{}\text{(15)}& \begin{array}{c}{\tilde{\text{P}}}_{\text{k}}=\left({\text{p}}_{\text{k}}^{\text{l}},{\text{ p}}_{\text{k}}^{\text{m}},{\text{ p}}_{\text{k}}^{\text{u}}\right)=\\ (\sqrt{{({\text{p}}_1^{\text{l}})}^2+{({\text{p}}_2^{\text{l}})}^2+\dots {\left({\text{p}}_{\text{k}}^{\text{l}}\right)}^2},\sqrt{{({\text{p}}_1^{\text{m}})}^2+{({\text{p}}_2^{\text{m}})}^2+\dots {\left({\text{p}}_{\text{k}}^{\text{m}}\right)}^2},\\ \sqrt{{({\text{p}}_1^{\text{u}})}^2+{({\text{p}}_2^{\text{u}})}^2+\dots {\left({\text{p}}_{\text{k}}^{\text{u}}\right)}^2})\end{array}\end{array}

\begin{array}{}\text{(16)}& \begin{array}{c}{\tilde{\text{P}}}_{\text{h}}=\left({\text{p}}_{\text{h}}^{\text{l}},{\text{ p}}_{\text{h}}^{\text{m}},{\text{ p}}_{\text{h}}^{\text{u}}\right)=\\ (\sqrt{{({\text{p}}_1^{\text{l}})}^2+{({\text{p}}_2^{\text{l}})}^2+\dots +{({\text{p}}_{\text{h}}^{\text{l}})}^2},\sqrt{{({\text{p}}_1^{\text{m}})}^2+{({\text{p}}_2^{\text{m}})}^2+\dots {\left({\text{p}}_{\text{h}}^{\text{m}}\right)}^2},\\ \sqrt{{({\text{p}}_1^{\text{u}})}^2+{({\text{p}}_2^{\text{u}})}^2+\dots {\left({\text{p}}_{\text{h}}^{\text{u}}\right)}^2})\end{array}\end{array}

Every alternative follows the same procedure.

\begin{array}{}\text{(17)}& \begin{array}{c}{\tilde{\text{R}}}_{\text{k}}=\left({\text{r}}_{\text{k}}^{\text{l}},{\text{ r}}_{\text{k}}^{\text{m}},{\text{ r}}_{\text{k}}^{\text{u}}\right)=\\ (\sqrt{{({\text{r}}_1^{\text{l}})}^2+{({\text{r}}_2^{\text{l}})}^2+\dots +{({\text{r}}_{\text{k}}^{\text{l}})}^2},\sqrt{{({\text{r}}_1^{\text{m}})}^2+{({\text{r}}_2^{\text{m}})}^2+\dots {\left({\text{r}}_{\text{k}}^{\text{m}}\right)}^2},\\ \sqrt{{({\text{r}}_1^{\text{u}})}^2+{({\text{r}}_2^{\text{u}})}^2+\dots {\left({\text{r}}_{\text{k}}^{\text{u}}\right)}^2})\end{array}\end{array}

\begin{array}{}\text{(18)}& \begin{array}{c}{\tilde{\text{R}}}_{\text{h}}=\left({\text{r}}_{\text{h}}^{\text{l}},{\text{ r}}_{\text{h}}^{\text{m}},{\text{ r}}_{\text{h}}^{\text{u}}\right)=\\ (\sqrt{{({\text{r}}_1^{\text{l}})}^2+{({\text{r}}_2^{\text{l}})}^2+\dots +{({\text{r}}_{\text{h}}^{\text{l}})}^2},\sqrt{{({\text{r}}_1^{\text{m}})}^2+{({\text{r}}_2^{\text{m}})}^2+\dots {\left({\text{r}}_{\text{h}}^{\text{m}}\right)}^2},\\ \sqrt{{({\text{r}}_1^{\text{u}})}^2+{({\text{r}}_2^{\text{u}})}^2+\dots {\left({\text{r}}_{\text{h}}^{\text{u}}\right)}^2})\end{array}\end{array}

Step 8. Two matrices \tilde{\text{Z}}\text{ and }{\tilde{\text{C}}}_{\text{i}} are created using equations (19) and (20). The first matrix is composed of optimal alternative, and the second matrix is composed of each alternative.

\begin{array}{}\text{(19)}& \tilde{\text{Z}}=\left[\begin{array}{cc}{\tilde{\text{P}}}_{\text{k}}& 0\\ 0& {\tilde{\text{P}}}_{\text{h}}\end{array}\right]\end{array}

\begin{array}{}\text{(20)}& {\tilde{\text{C}}}_{\text{i}}=\left[\begin{array}{cc}{\tilde{\text{R}}}_{\text{ik}}& 0\\ 0& {\tilde{\text{R}}}_{\text{ih}}\end{array}\right]\end{array}

Step 9: \tilde{\text{Z}}\text{ and }{\tilde{\text{C}}}_{\text{i}} Matrices are multiplied to obtain {\tilde{\text{X}}}_{\text{i}} matrix using equation (21).

\begin{array}{}\text{(21)}& {\tilde{\text{X}}}_{\text{i}}=\tilde{\text{Z}}\times {\tilde{\text{C}}}_{\text{i}}=\left[\begin{array}{cc}{\tilde{\text{x}}}_{11;\text{i}}& 0\\ 0& {\tilde{\text{x}}}_{22;\text{i}}\end{array}\right]\end{array}

Step 10.1: Ranking of the enablers by using FMCRAT.

The fuzzy trace of the matrix {\tilde{\text{X}}}_{\text{i}} is computed by using equation (22).

\begin{array}{}\text{(22)}& \text{tr}\left({\tilde{\text{X}}}_{\text{i}}\right)={\tilde{\text{x}}}_{11;\text{i}}+{\tilde{\text{x}}}_{22;\text{i}}=\left({\text{x}}_{11,\text{i}}^{\text{l}}+{\text{x}}_{22,\text{i}}^{\text{l}},{\text{ x}}_{11,\text{i}}^{\text{m}}+{\text{x}}_{22,\text{i}}^{\text{m}},{\text{ x}}_{11,\text{i}}^{\text{u}}+{\text{x}}_{22,\text{i}}^{\text{u}}\right)\end{array}

In the above equation (22), trace is given in fuzzy numbers, and this value is defuzzified by using ({\text{X}}_{\text{i}})=\frac{{\text{X}}_{\text{i}}^{\text{l}}+4{\text{X}}_{\text{i}}^{\text{m}}+{\text{X}}_{\text{i}}^{\text{u}}}{6}. Now the best alternative is the one with the highest tr (X_i).

Step 10.2: Ranking of the enablers by using FRAPS

The fuzzy perimeter of the right-angle triangle represents the optimal alternative’s perimeter computed by equation (23). The base and perpendicular side of this triangle, in terms of triangular fuzzy numbers, are represented by components \left({\text{p}}_{\text{k}}^{\text{l}},{\text{ p}}_{\text{k}}^{\text{m}},{\text{ p}}_{\text{k}}^{\text{u}}\right) and ({\text{p}}_{\text{h}}^{\text{l}}{\text{, p}}_{\text{h}}^{\text{m}}{\text{, p}}_{\text{h}}^{\text{u}}), respectively.

\begin{array}{}\text{(23)}& \begin{array}{c}\tilde{\text{S}}=\left({\text{S}}^{\text{l}},{\text{ S}}^{\text{m}},{\text{ S}}^{\text{u}}\right)=\\ (\left({\text{p}}_{\text{k}}^{\text{l}}+{\text{p}}_{\text{h}}^{\text{l}}+\sqrt{{\text{p}}_{\text{k}}^{\text{l}}{}^2+{\text{p}}_{\text{h}}^{\text{l}}{}^2}\right),\left({\text{p}}_{\text{k}}^{\text{m}}+{\text{p}}_{\text{h}}^{\text{m}}+\sqrt{{\text{p}}_{\text{k}}^{\text{m}}{}^2+{\text{p}}_{\text{h}}^{\text{m}}{}^2}\right),\\ \left({\text{p}}_{\text{k}}^{\text{u}}+{\text{p}}_{\text{h}}^{\text{u}}+\sqrt{{\text{p}}_{\text{k}}^{\text{u}}{}^2+{\text{p}}_{\text{h}}^{\text{u}}{}^2}\right))\end{array}\end{array}

The fuzzy perimeter of each alternative is calculated by the same method using equation (24).

\begin{array}{}\text{(24)}& \begin{array}{c}{\tilde{\text{S}}}_{\text{e}}=\left({\text{S}}_{\text{e}}^{\text{l}}{,}_{ }{\text{S}}_{\text{e}}^{\text{m}}{,}_{ }{\text{S}}_{\text{e}}^{\text{u}}\right)=\\ (\left({\text{r}}_{\text{k}}^{\text{l}}+{\text{r}}_{\text{h}}^{\text{l}}+\sqrt{{\text{r}}_{\text{k}}^{\text{l}}{}^2+{\text{r}}_{\text{h}}^{\text{l}}{}^2}\right), \left({\text{r}}_{\text{k}}^{\text{m}}+{\text{r}}_{\text{h}}^{\text{m}}+\sqrt{{\text{r}}_{\text{k}}^{\text{m}}{}^2+{\text{r}}_{\text{h}}^{\text{m}}{}^2}\right), \\ \left({\text{r}}_{\text{k}}^{\text{u}}+{\text{r}}_{\text{h}}^{\text{u}}+\sqrt{{\text{r}}_{\text{k}}^{\text{u}}{}^2+{\text{r}}_{\text{h}}^{\text{u}}{}^2}\right))\end{array}\end{array}

The ratio of the fuzzy perimeter of each alternative to the fuzzy perimeter of the optimal alternative is known as the fuzzy perimeter similarity {\widetilde{\text{PS}}}_{\text{e}}, which is computed using equation (25).

\begin{array}{}\text{(25)}& {\widetilde{\text{PS}}}_{\text{e}}=\left({\text{PS}}_{\text{e}}^{\text{l}},{\text{ PS}}_{\text{e}}^{\text{m}},{\text{ PS}}_{\text{e}}^{\text{u}}\right)=\left(\frac{{\text{S}}_{\text{e}}^{\text{l}}}{{\text{s}}^{\text{u}}},\frac{{\text{S}}_{\text{e}}^{\text{m}}}{{\text{s}}^{\text{m}}},\frac{{\text{S}}_{\text{e}}^{\text{u}}}{{\text{s}}^{\text{l}}}\right)\end{array}

The perimeter similarity values were defuzzified using PS_e=\frac{\left({\text{PS}}_{\text{e}}^{\text{l}}+4\times {\text{PS}}_{\text{e}}^{\text{m}}+{\text{PS}}_{\text{e}}^{\text{u}}\right)}{6} and the alternatives were ranked in descending order based on the defuzzified perimeter similarity.

4 Data Collection, Results, and Validation of Results

4.1 Data collection

Google Forms was used to create a structured questionnaire and was sent to five domain experts via email who met the relevant qualifications and experience required (see Table 3), and involved two different sections: the first section included pairwise comparisons among the evaluation criteria, and the second included the expert rating of the healthcare enablers of ChatGPT relative to the evaluation criteria, using a fuzzy linguistic scale as described in Sections 3.1 and 3.2, and the resulting pairwise comparisons and individual ratings are shown in Tables A1–A5 and A6–A10 respectively (see annexure).

4.2 Results

Based on these expert ratings, the criteria weights were determined, and the enablers were subsequently ranked using the FMCRAT and FRAPS methods. Detailed calculation procedures are presented in sections 4.2.1 and 4.2.2.

4.2.1 Calculations of fuzzy weight of evaluation criteria

Initially, a decision matrix for the criteria versus criteria comparisons is developed by converting the linguistic scale into TFNs, as outlined in Table 4. The individual ratings provided by the five experts are detailed in Tables A1–A5 (see annexure). The initial fuzzy decision matrix for the criteria versus criteria comparisons, based on the ratings from expert 1, is presented in Table A11 (see annexure). Subsequently, the fuzzy geometric mean is computed using equation (2), as shown in Table A12 (see annexure). The fuzzy weights of the criteria, derived from this fuzzy geometric mean using equation (3), are detailed in Table A13 (see annexure). The same methodology is applied to compute the fuzzy weights based on the ratings from the remaining experts. Table 6 presents the fuzzy weights derived from the ratings of all five experts.

After calculating the fuzzy weights from all expert reviews, the average fuzzy weight is computed. Then, it is converted to a crisp weight using equation (4). Table 7 shows both the average fuzzy weights and the crisp weights.

In this study, risk (C2) is identified as a non-beneficial criterion, as the goal is to minimise its effect. In contrast, reliability (C1), data privacy (C3), sustainability (C4), cost effectiveness (C5), problem solving (C6), accuracy (C7), availability (C8), and collaboration (C9) are considered beneficial criteria, as the goal is to maximise its effect. The findings reveal that risk (C2) holds the highest crisp weight, underscoring its critical importance in the evaluation process. Conversely, collaboration (C9) has the lowest crisp weight, indicating it is the least influential criterion. Figure 3 presents the relative importance of the criteria.

4.2.2 Ranking of enablers by using FMCRAT and FRAPS.

To rank the enablers, this study uses the FMCRAT and FRAPS methods. Initially, the linguistic scales provided by experts, as shown in Tables A6–A10 (see annexure), for comparing criteria with ChatGPT healthcare enablers are converted into TFNs using the scale presents in Table 5. Subsequently, an initial fuzzy decision matrix for the criteria versus enablers comparisons is created by calculating the arithmetic mean of the experts’ ratings. The resulting initial decision matrix is displayed in Table A14 (see annexure). After constructing the initial decision matrix, the normalised fuzzy decision matrix is computed using equations (6) and (7) and is presented in Table A15 (see annexure).

Following the normalisation, the fuzzy weighted normalised matrix is calculated using equation (8), with the results shown in Table A16 (see annexure). The optimal alternative (enabler) is then identified using equations (9) and (10), with the corresponding results presented in Table A17 (see annexure).

Using equations (11)–(18), the magnitudes of the fuzzy optimal alternative and all alternatives are computed, with the results presented in Table A18 (see annexure). Subsequently, the trace of the matrix is calculated as a fuzzy number using equations (19)–(22) and then defuzzified to obtain crisp trace values. Table 8 presents the fuzzy traces, defuzzified traces, and the corresponding rankings of the enablers. Figure 4 illustrates the relative importance of the ChatGPT healthcare enablers as determined by the FMCRAT method.

The enablers are ranked using the FRAPS method. Initially, the fuzzy perimeter and fuzzy perimeter similarity are calculated through equations (23)–(25). These fuzzy perimeter similarity values are then defuzzified, allowing the enablers to be ranked in descending order based on the defuzzified perimeter similarity. Table 9 displays the ranking of the enablers as determined by the FRAPS method. Figure 5 depicts the relative importance of the ChatGPT healthcare enablers as assessed by the FRAPS method.

4.3 Validation of Results

Sensitivity analysis (SA) was employed in this study to ensure the robustness and reliability of the findings derived using the FMCRAT and FRAPS methodologies. In MCDM, SA is important in decision-making. It checks how changes in input data or weights affect the results. This helps to see if the results are reliable even when there is uncertainty [78]. A total of 36 combinations were evaluated by interchanging the weights of any two of the nine criteria (2 out of 9) in each trial. For the FMCRAT approach, the enabler ranking was determined as follows: E9 > E4 > E8 > E2 > E7 > E3 > E6 > E5 > E1. Through several iterations, E1, E2, E5, and E6 consistently held the same rank 33 times, E3 and E7 maintained their rank 32 times, E4 held its rank 35 times, E8 remained in the same rank 33 times, and E9 remained in the same rank 34 times. In contrast, the FRAPS method exhibited a highly consistent ranking sequence, with E9 securing the top position 35 times. E4 and E8 were ranked second and third, respectively, each appearing 33 times. E2 was ranked fourth 32 times, E7 was fifth 33 times, E3 was sixth 34 times, E6 was seventh 33 times, E5 was eighth 34 times, and E1 was ninth 32 times. Overall, SA indicates that ranking stability is high for both the FMCRAT and FRAPS methods, underscoring the robustness and reliability of the proposed decision-making model. The SA generally confirms that both the FMCRAT and FRAPS methods exhibit high stability in their rankings, demonstrating the robustness and reliability of the proposed decision-making framework. Figures 6 and 7 illustrate the SA results for the FMCRAT and FRAPS methods, respectively.

5 Discussions

The prioritisation outcomes of the FMCRAT and FRAPS methods provide valuable insights into the relative importance of ChatGPT enablers in developing sustainable healthcare systems. The defuzzified trace and the similarity of the values in the perimeter are the highest for Patient Feedback (E9), which highlights the quality of patient-centric insights in enhancing service quality and facilitating continuous system development. This finding aligns with the broader emphasis on patient-centric care observed in Medical 4.0 studies, where patient involvement and feedback mechanisms are recognised as critical enablers of healthcare transformation. The second and third-ranked enablers, Clinical Decision Support (E4) and Medical Queries (E8), note the ability of ChatGPT to support evidence-based decision-making and to provide fast access to trusted medical information. These enablers are similar to the significance of AI and better diagnostics in Medical 4.0 frameworks, which show how conversational AI can reinforce clinical intelligence and assist healthcare professionals in the decision-making process. Healthcare Education (E2) and Drug Information (E7) also underscore the growing role of ChatGPT in knowledge sharing, professional education, and safe medication use, which will help sustain the system in the long term through capacity building. The mid-ranked enablers, Appropriate Treatment (E3), Fraud Detection (E6), and Mental Health Support (E5), reflect areas where ChatGPT has emerging but still developing potential. These enablers correspond to treatment optimisation, data integrity, and patient well-being, which are also acknowledged in Medical 4.0 as important but evolving domains requiring further technological maturity and integration. Interestingly, Telemedicine (E1) is the lowest of the identified enablers. This does not mean it is less important. Still, it implies that telemedicine is already a proven part of modern healthcare systems, with ChatGPT serving as an additive and supporting layer rather than a major driver.

6 Implications of this Research

The findings of this study have significant practice, policy, and healthcare provision implications regarding the need to prioritise ChatGPT enablers in the development of sustainable healthcare systems using a FMCDM methodology. The study provides a systematic framework for ranking the identified enablers using FMCRAT and FRAPS, offering policymakers, hospital administrators, and healthcare planners a structured foundation for strategic decision-making regarding the implementation of ChatGPT in healthcare operations. These prioritisation insights can be effectively used to allocate financial, technological, and human resources to the most influential areas to maximise the impact of AI adoption whilst ensuring long-term system sustainability. Within the healthcare practice context, the findings provide insights into how ChatGPT can be used to enhance patient-centric care through patient feedback, clinical decision support, medical query processing, healthcare education, and drug information services. This facilitates improved service quality, enhanced clinical efficiency, reduced information asymmetry, and better patient engagement. The prioritised enablers also guide healthcare organisations in identifying where ChatGPT can be most effectively embedded within clinical workflows, administrative processes, training programs, and patient interaction channels. This targeted integration can help overcome many typical implementation barriers, facilitate technology adoption among healthcare professionals, and enable easier shifts to AI-assisted healthcare delivery. Regarding the resilience of the healthcare system, the study shows that prioritising the most important ChatGPT enablers can help create adaptive, responsive, and knowledge-based healthcare settings that meet changing patient demands and address issues in healthcare operations. The framework supports evidence-based planning for digital transformation initiatives aligned with sustainability goals, particularly in improving accessibility, quality of care, transparency, and continuous system improvement. In terms of academic value, this research serves a validating role by confirming the existence of FMCDM methods for AI enabler prioritisation in complex healthcare environments. It offers a methodological underpinning that can be replicated in further research on AI-enabled sustainable healthcare development.

7 Limitations and Future Scope of this Research

Despite offering meaningful insights into the prioritisation of ChatGPT enablers for sustainable healthcare systems, this study has certain limitations. First, the analysis relies constantly on expert judgment. Although the uncertainty was addressed and individual bias minimised using fuzzy logic, there remains the risk of residual subjectivity. Second, the limited number of experts that participated and the utilisation of an already identified collection of enablers might limit the breadth, representativeness, and generalisability of the results. Third, the research is more conceptual and methodological, with results not confirmed using empirical or real-world data on healthcare performance. To overcome these limitations, future studies can consider a broader group of stakeholders, including clinicians, healthcare administrators, policymakers, AI practitioners, and patients, to gain a more comprehensive perspective. Further refinement of the set of enablers with the help of empirical exploration would also strengthen the framework. The practical applicability of the proposed model will be reinforced by using it in real healthcare system and supporting it with quantitative performance indicators. Also, further studies on the sustainability and scalability of AI-enabled interventions can be conducted through comparative analyses with other AI models, as well as longitudinal studies across various healthcare settings. These extensions will improve the suggested framework and make it more applicable to developing sustainable AI-driven healthcare systems.

8 Conclusions

This study establishes a structured, uncertainty-aware decision-support framework for evaluating and prioritising ChatGPT enablers in sustainable healthcare systems, integrating FMCRAT and FRAPS. The results demonstrate that Patient Feedback, Clinical Decision Support, and Medical Queries are the most influential enablers, highlighting ChatGPT’s capacity to enhance clinical accuracy, strengthen patient engagement, and improve service efficiency three core pillars of healthcare sustainability. From a practical perspective, these findings provide clear guidance for healthcare policymakers, administrators, and system developers by indicating where strategic investments, regulatory focus, and technological optimisation efforts should be concentrated to maximise long-term value and system resilience. The robustness of prioritisation, as verified by a sensitivity analysis, supports the soundness of the suggested framework for making real-life decisions under uncertainty. In addition to the managerial timeliness of the topic, this research paper contributes to the use of FMCDM methods in digital healthcare by providing a robust methodology for evaluating emerging AI-related technologies as a feature of sustainability-based frameworks. This study can be advanced in future research through empirical validation across various healthcare environments, broader involvement of experts, and active reassessment of enablers as conversational artificial intelligence technologies and medical ecosystems continue to advance.

References

[1] E.-R. Youssef, M. Meriem, and H. Ait-Lemqeddem, “ChatGPT and ethics in healthcare facilities: an overview and innovations in technical efficiency analysis,” AI and Ethics 2025 5:4, vol. 5, no. 4, pp. 3479–3496, Feb. 2025,

[2] N. Rane, “Roles and Challenges of ChatGPT and Similar Generative Artificial Intelligence for Achieving the Sustainable Development Goals (SDGs),” SSRN Electronic Journal, Aug. 2023,

[3] O. A. AlJaberi, M. Hussain, and P. R. Drake, “A framework for measuring sustainability in healthcare systems,” Int. J. Healthc. Manag., vol. 13, no. 4, pp. 276–285, Oct. 2020,

[4] M. Faezipour and S. Ferreira, “Applying systems thinking to assess sustainability in healthcare system of systems,” International Journal of System of Systems Engineering, vol. 2, no. 4, pp. 290–308, 2011,

[5] R. Chaurasiya and D. Jain, “Hybrid MCDM method on pythagorean fuzzy set and its application,” Decision Making: Applications in Management and Engineering, vol. 6, no. 1, pp. 379–398, Apr. 2023,

[6] Y. Ommane, M. Lachgar, M. Hanine, M. D. Laanaoui, and S. Kaouri, “Optimizing healthcare supply chain capacity planning during disasters using fuzzy AHP and TOPSIS methods,” Scientific Reports 2025 15:1, vol. 15, no. 1, pp. 35335–, Oct. 2025,

[7] Y. Ojo, O. A. Makinde, O. V. Babatunde, G. Babatunde, and S. Okeowo, “Evaluating AI-Driven Mental Health Solutions: A Hybrid Fuzzy Multi-Criteria Decision-Making Approach,” AI 2025, Vol. 6, Page 14, vol. 6, no. 1, p. 14, Jan. 2025,

[8] R. Kumaret al., “Fuzzy-Based Symmetrical Multi-Criteria Decision-Making Procedure for Evaluating the Impact of Harmful Factors of Healthcare Information Security,” Symmetry 2020, Vol. 12, Page 664, vol. 12, no. 4, p. 664, Apr. 2020,

[9] Q. Wei, Z. Yao, Y. Cui, B. Wei, Z. Jin, and X. Xu, “Evaluation of ChatGPT-generated medical responses: A systematic review and meta-analysis,” J. Biomed. Inform., vol. 151, Mar. 2024,

[10] A. Valencia-Ariaset al., “Understanding telemedicine adoption: evidence, gaps, and future perspectives for sustainable healthcare,” Cogent Soc. Sci., vol. 10, no. 1, p. 2306712, Dec. 2024,

[11] Dr. A. S. George and A. S. H. George, “Telemedicine: A New Way to Provide Healthcare,” Partners Universal International Innovation Journal, vol. 1, no. 3, pp. 98–129, Jun. 2023,

[12] M. Stoltzfus, A. Kaur, A. Chawla, V. Gupta, F. N. U. Anamika, and R. Jain, “The role of telemedicine in healthcare: an overview and update,” The Egyptian Journal of Internal Medicine 2023 35:1, vol. 35, no. 1, pp. 1–5, Jun. 2023,

[13] M. X. Jin, S. Y. Kim, L. J. Miller, G. Behari, and R. Correa, “Telemedicine: Current Impact on the Future,” 2020,

[14] A. Teherani, H. Nishimura, L. Apatira, T. Newman, and S. Ryan, “Identification of core objectives for teaching sustainable healthcare education,” Med. Educ. Online, vol. 22, no. 1, Jan. 2017,

[15] A. Costelloet al., “Managing the health effects of climate change. Lancet and University College London Institute for Global Health Commission,” The Lancet, vol. 373, no. 9676, pp. 1693–1733, 2009,

[16] “Correction to: AMEE Consensus Statement: Planetary health and education for sustainable healthcare (Medical Teacher, 43, 3, (272–286), 10.1080/0142159X.2020.1860207),” Med. Teach., vol. 44, no. 11, p. (I), 2022,

[17] E. Shawet al., “AMEE Consensus Statement: Planetary health and education for sustainable healthcare,” Med. Teach., vol. 43, no. 3, pp. 272–286, 2021,

[18] G. Brandet al., “‘I teach it because it is the biggest threat to health’: Integrating sustainable healthcare into health professions education,” Med. Teach., vol. 43, no. 3, pp. 325–333, 2021,

[19] M. Romanelli, “Towards Sustainable Health Care Organizations,” Management Dynamics in the Knowledge Economy, vol. 5, no. 3, pp. 377–394, 2017.

[20] M. Tung and S. Kaur, “Sustainable Healthcare Systems,” Primary Care Revisited: Interdisciplinary Perspectives for a New Era, pp. 51–66, Jan. 2020,

[21] J. S. Heerink, R. Oudega, R. Hopstaken, H. Koffijberg, and R. Kusters, “Clinical decision rules in primary care: necessary investments for sustainable healthcare,” Prim. Health Care Res. Dev., vol. 24, no. e34, p. e34, May 2023,

[22] S. Wieringa, H. McGuire, Q. Wang, P. Van Der Wees, and B. Shaw, “Making sustainable healthcare decisions: three turns towards sustainable guidelines,” BMJ Evid. Based. Med., vol. 29, no. 4, pp. 219–222, Aug. 2024,

[23] N. Tutticci and N. M. Huss, “An analysis of sustainable decision-making using clinical reasoning,” Teaching and Learning in Nursing, vol. 20, no. 2, pp. 131–136, Apr. 2025,

[24] C. Marzorati and G. Pravettoni, “Value as the key concept in the health care system: How it has influenced medical practice and clinical decision-making processes,” J. Multidiscip. Healthc., vol. 10, pp. 101–106, 2017,

[25] E. Bohlmeijer and G. Westerhof, “The Model for Sustainable Mental Health: Future Directions for Integrating Positive Psychology Into Mental Health Care,” Front. Psychol., vol. 12, p. 747999, Oct. 2021,

[26] I. Qureshiet al., “Caring for Those Who Take Care of Others: Developing Systemic and Sustainable Mental Health Support for the Diverse Healthcare Workforce in the United Kingdom,” International Journal of Environmental Research and Public Health 2023, Vol. 20, Page 3242, vol. 20, no. 4, p. 3242, Feb. 2023,

[27] V. Barik, M. D. Samsuddin, P. Khamkat, and B. B. Barik, “Mental Health and Well-Being in Sustainable Practices,” World Sustainability Series, vol. Part F3017, pp. 169–179, 2024,

[28] G. Thornicroft and V. Patel, “Including mental health among the new sustainable development goals,” BMJ, vol. 349, Aug. 2014,

[29] A. A. Amponsah, A. F. Adekoya, and B. A. Weyori, “A novel fraud detection and prevention method for healthcare claim processing using machine learning and blockchain technology,” Decision Analytics Journal, vol. 4, p. 100122, Sep. 2022,

[30] J. Lu, K. Lin, R. Chen, M. Lin, X. Chen, and P. Lu, “Health insurance fraud detection by using an attributed heterogeneous information network with a hierarchical attention mechanism,” BMC Med. Inform. Decis. Mak., vol. 23, no. 1, pp. 1–17, Dec. 2023,

[31] R. Nurul Fatimah, R. Adlia Syakurah, A. Kesehatan Nasional, S. kesehatan, and P. Kesehatan, “Potential Fraud in The Implementation of National Health Insurance in The Health Sector: Systematic Review,” JMMR (Jurnal Medicoeticolegal dan Manajemen Rumah Sakit), vol. 10, no. 3, pp. 255–270, Nov. 2021,

[32] A. M. Keerthi, S. Ramapriya, S. B. Kashyap, P. K. Gupta, and B. S. Rekha, “Pharmaceutical Management Information Systems: A Sustainable Computing Paradigm in the Pharmaceutical Industry and Public Health Management,” EAI/Springer Innovations in Communication and Computing, pp. 33–51, 2021,

[33] P. K. Mukherjee, S. Banerjee, C. K. Katiyar, S. Sharma, and N. Chattopadhyay, “Traditional Medical System (TMS) for Sustainable Healthcare in India,” Drug Discovery and Drug Development: The Indian Narrative, pp. 1–36, Jan. 2021,

[34] H. ten Have and M. do C. Patrão Neves, “Traditional Medicine,” Dictionary of Global Bioethics, pp. 1007–1008, 2021,

[35] S. Biswas, “ChatGPT and the Future of Medical Writing,” https://doi.org/10.1148/radiol.223312, vol. 307, no. 2, Feb. 2023,

[36] A. Alfouri and S. Alfouri, “Artificial Intelligence in Medicine and Sustainable Development.,” Veterinary Medicine & Public Health Journal, vol. 6, no. 3, p. 105, Sep. 2025,

[37] J. Rapin, J. Pellet, C. Mabire, S. Gendron, and C. A. Dubois, “How does feedback shared with interprofessional health care teams shape nursing performance improvement systems? A rapid realist review protocol,” Syst. Rev., vol. 8, no. 1, pp. 1–10, Jul. 2019,

[38] J. Greenhalghet al., “HEALTH SERVICES AND DELIVERY RESEARCH Functionality and feedback: a realist synthesis of the collation, interpretation and utilisation of patient-reported outcome measures data to improve patient care,” vol. 5, no. 2, pp. 2050–4349, 2017,

[39] A. Coulter, L. Locock, S. Ziebland, and J. Calabrese, “Collecting data on patient experience is not enough: they must be used to improve care,” BMJ, vol. 348, Mar. 2014,

[40] R. Eslamipoor and A. Nobari, “A reliable and sustainable design of supply chain in healthcare under uncertainty regarding environmental impacts,” Journal of Applied Research on Industrial Engineering, vol. 10, no. 2, pp. 256–272, Jun. 2023,

[41] A. Aqil, K. Saldana, N. uddin Mian, and M. Ndu, “Reliability and validity of an innovative high performing healthcare system assessment tool,” BMC Health Serv. Res., vol. 23, no. 1, pp. 1–18, Dec. 2023,

[42] C. Corvalanet al., “Towards Climate Resilient and Environmentally Sustainable Health Care Facilities,” International Journal of Environmental Research and Public Health 2020, Vol. 17, Page 8849, vol. 17, no. 23, p. 8849, Nov. 2020,

[43] G. Russo, A. M. Tartaglione, and Y. Cavacece, “Empowering Patients to Co-Create a Sustainable Healthcare Value,” Sustainability 2019, Vol. 11, Page 1315, vol. 11, no. 5, p. 1315, Mar. 2019,

[44] J. Iwelunmoret al., “Toward the sustainability of health interventions implemented in sub-Saharan Africa: A systematic reviewand conceptual framework,” Implementation Science, vol. 11, no. 1, pp. 1–27, Mar. 2016,

[45] A. Jovanovićet al., “Assessing resilience of healthcare infrastructure exposed to COVID-19: emerging risks, resilience indicators, interdependencies and international standards,” Environ. Syst. Decis., vol. 40, no. 2, pp. 252–286, Jun. 2020,

[46] N. S. Godbole and J. Lamb, “The triple challenge for the healthcare industry: Sustainability, privacy, and cloud-centric regulatory compliance,” 2013 10th International Conference and Expo on Emerging Technologies for a Smarter World, CEWIT 2013, 2013,

[47] V. Liu, L. May, W. Caelli, and P. Croll, “A Sustainable Approach to Security and Privacy in Health Information Systems,” Proceedings of the 18th Australasian Conference on Information Systems, 2007.

[48] A. Pereno and D. Eriksson, “A multi-stakeholder perspective on sustainable healthcare: From 2030 onwards,” Futures, vol. 122, p. 102605, Sep. 2020,

[49] Y. Zurynskiet al., “How can the healthcare system deliver sustainable performance? A scoping review,” BMJ Open, vol. 12, no. 5, p. e059207, May 2022,

[50] J. Braithwaiteet al., “Built to last? The sustainability of healthcare system improvements, programmes and interventions: a systematic integrative review,” BMJ Open, vol. 10, no. 6, p. e036453, Jun. 2020,

[51] J. Piera-Jiménez and F. Lupiáñez-Villanueva, “Decision making in digital health: usage of the cost-effectiveness analysis to improve the sustainability of the healthcare system,” Nov. 19, 2021, Universitat Oberta de Catalunya (UOC). Accessed: Oct. 22, 2025. [Online]. Available: https://hdl.handle.net/10609/143266

[52] “Determining Strategies for Cost Effectiveness in Sustainable Healthcare Delivery – An Emerging Market Study - ProQuest.” Accessed: Oct. 22, 2025. [Online]. Available: https://www.proquest.com/openview/91d126f4294665b9091f90643cc66612/1?cbl=18750&diss=y&pq-origsite=gscholar

[53] A. Erjaee, S. Hendiani, S. Moradi, and M. Bagherpour, “Healthcare Sustainability Evaluation Using a Hybrid Fuzzy Multi-Criteria Decision-Making Model,” International Journal of Fuzzy Systems, vol. 24, no. 2, pp. 1182–1202, Mar. 2022,

[54] F. Alves, L. A. Costa, A. M. A. C. Rocha, A. I. Pereira, and P. Leitão, “The Sustainable Home Health Care Process Based on Multi-Criteria Decision-Support,” Mathematics 2023, Vol. 11, Page 6, vol. 11, no. 1, p. 6, Dec. 2022,

[55] N. M. Mazar, M. H. Ehrampoush, A. A. Ebrahimi, M. T. Ghaneian, M. Dolatabadi, and R. Askari, “Investigatinon of Green Hospital Standards in Jiroft Hospitals,” Journal of Environmental Health and Sustainable Development, vol. 4, no. 1, pp. 701–709, Mar. 2019,

[56] L. P. Vishwakarma, R. K. Singh, R. Mishra, and A. Kumari, “Application of artificial intelligence for resilient and sustainable healthcare system: systematic literature review and future research directions,” Int. J. Prod. Res., vol. 63, no. 2, pp. 822–844, Jan. 2025,

[57] J. Barjis, G. Kolfschoten, and J. Maritz, “A sustainable and affordable support system for rural healthcare delivery,” Decis. Support Syst., vol. 56, no. 1, pp. 223–233, Dec. 2013,

[58] F. Uras, “Sustainable healthcare and medical laboratories: The impact of global collaborations between frameworks and initiatives,” Clin. Biochem., vol. 138, p. 110945, Aug. 2025,

[59] S. C. Walpole and F. Mortimer, “Evaluation of a collaborative project to develop sustainable healthcare education in eight UK medical schools,” Public Health, vol. 150, pp. 134–148, Sep. 2017,

[60] T. Maghsoudi, R. Cascón-Pereira, and A. B. H. Lara, “The Role of Collaborative Healthcare in Improving Social Sustainability: A Conceptual Framework,” Sustainability 2020, Vol. 12, Page 3195, vol. 12, no. 8, p. 3195, Apr. 2020,

[61] M. Ortíz-Barrios, N. Jaramillo-Rueda, M. Gul, M. Yucesan, G. Jiménez-Delgado, and J. J. Alfaro-Saíz, “A Fuzzy Hybrid MCDM Approach for Assessing the Emergency Department Performance during the COVID-19 Outbreak,” International Journal of Environmental Research and Public Health 2023, Vol. 20, Page 4591, vol. 20, no. 5, p. 4591, Mar. 2023,

[62] A. A. Al Mohamed, S. Al Mohamed, and M. Zino, “Application of fuzzy multicriteria decision-making model in selecting pandemic hospital site,” Future Business Journal 2023 9:1, vol. 9, no. 1, pp. 14-, Apr. 2023,

[63] A. Mardani, A. Jusoh, and E. K. Zavadskas, “Fuzzy multiple criteria decision-making techniques and applications – Two decades review from 1994 to 2014,” Expert Syst. Appl., vol. 42, no. 8, pp. 4126–4148, May 2015,

[64] L. A. Zadeh, “Fuzzy sets as a basis for a theory of possibility,” Fuzzy Sets Syst., vol. 1, no. 1, pp. 3–28, Jan. 1978,

[65] A. A. Makki and R. M. S. Abdulaal, “A Hybrid MCDM Approach Based on Fuzzy MEREC-G and Fuzzy RATMI,” Mathematics, vol. 11, no. 17, p. 3773, Sep. 2023,

[66] J. lu Shi and W. hsiang Lai, “Fuzzy AHP approach to evaluate incentive factors of high-tech talent agglomeration,” Expert Syst. Appl., vol. 212, p. 118652, Feb. 2023,

[67] S. Salari, M. Sadeghi-Yarandi, and F. Golbabaei, “An integrated approach to occupational health risk assessment of manufacturing nanomaterials using Pythagorean Fuzzy AHP and Fuzzy Inference System,” Scientific Reports 2024 14:1, vol. 14, no. 1, pp. 180-, Jan. 2024,

[68] T. L. Saaty, “Decision making with the Analytic Hierarchy Process,” Scientia Iranica, vol. 9, no. 3, pp. 215–229, 2002,

[69] O. S. Vaidya and S. Kumar, “Analytic hierarchy process: An overview of applications,” Eur. J. Oper. Res., vol. 169, no. 1, pp. 1–29, Feb. 2006,

[70] Y. M. Wang, Y. Luo, and Z. Hua, “On the extent analysis method for fuzzy AHP and its applications,” Eur. J. Oper. Res., vol. 186, no. 2, pp. 735–747, Apr. 2008,

[71] M. N. Akhtar, A. Haleem, M. Javaid, and M. Vasif, “Understanding Medical 4.0 implementation through enablers: An integrated multi-criteria decision-making approach,” Informatics and Health, vol. 1, no. 1, pp. 29–39, Mar. 2024,

[72] Z. Ayaǧ and R. G. Özdemir, “A fuzzy AHP approach to evaluating machine tool alternatives,” J. Intell. Manuf., vol. 17, no. 2, pp. 179–190, Apr. 2006,

[73] K. Urošević, Z. Gligorić, I. Miljanović, C. Beljić, and M. Gligorić, “Novel Methods in Multiple Criteria Decision-Making Process (MCRAT and RAPS)—Application in the Mining Industry,” Mathematics 2021, Vol. 9, Page 1980, vol. 9, no. 16, p. 1980, Aug. 2021,

[74] R. M. S. Abdulaal and O. A. Bafail, “Two New Approaches (RAMS-RATMI) in Multi-Criteria Decision-Making Tactics,” Journal of Mathematics, vol. 2022, no. 1, p. 6725318, Jan. 2022,

[75] A. Ulutaş, F. Balo, and A. Topal, “Identifying the Most Efficient Natural Fibre for Common Commercial Building Insulation Materials with an Integrated PSI, MEREC, LOPCOW and MCRAT Model,” Polymers 2023, Vol. 15, Page 1500, vol. 15, no. 6, p. 1500, Mar. 2023,

[76] A. Ulutaş, A. Topal, D. Karabasevic, and F. Balo, “Selection of a Forklift for a Cargo Company with Fuzzy BWM and Fuzzy MCRAT Methods,” Axioms 2023, Vol. 12, Page 467, vol. 12, no. 5, p. 467, May 2023,

[77] A. Puška, Ž. Stević, and I. Stojanović, “Selection of Sustainable Suppliers Using the Fuzzy MARCOS Method,” Current Chinese Science, vol. 1, no. 2, pp. 218–229, Feb. 2021,

[78] W. Wolters, B. M.-E. J. of O. Research, and undefined 1995, “Novel types of sensitivity analysis for additive MCDM methods,” ElsevierWTM Wolters, B MareschalEuropean Journal of Operational Research, 1995•Elsevier, Accessed: Oct. 28, 2025. [Online]. Available: https://www.sciencedirect.com/science/article/pii/0377221793E0343V

RIGHTS & PERMISSIONS

Higher Education Press 2026