DECISION SUPPORT IN HEALTH, HOW TO INCREASE THE USE OF MACHINE LEARNING? AN APPROACH IN NEPHROLOGY.

Ciências Exatas e da Terra, Volume 27 - Edição 128/NOV 2023 / 24/11/2023

REGISTRO DOI: 10.5281/zenodo.10157808

Antonio Teixeira Neto1
Natasha Correia Queiroz Lino2
Leandro Carlos de Souza3

Abstract. Chronic kidney disease is responsible for approximately 1.2 million deaths each year worldwide. The disease is associated with high health costs and a significant reduction in the patients’ quality of life. Early diagnosis is essential to prevent and delay its effects. The use of technologies to support medical decisions becomes essential to optimize the resources of the public health system and improve the patients’ quality of life. The result of this research presents the advance in the use of machine learning for decision support in health, presenting the main publications of articles in this area.

1. Introduction

Renal treatment presents a diversity of results, which can be explained by variability in biology, access to care, environmental factors, and differences in the health system Levin (2008). As reported by Gomes (2018) it is a disease with an average global prevalence of 12% to 15%, strongly associated with excessive health costs, high medication burden triggering a significant loss of patients’ quality of life. CKD rose from 36th to 12th position in the global burden of disease ranking (Global Burden of Disease – GBD-Study) in the last two decades Lunney et al. (2018) and it is estimated that 956,200 people died from CKD, an increase of 134% compared to 1990, one of the largest increases among the main causes of death Ene-Iordache et al. (2016). It is currently estimated that 1.2 million people have died from chronic kidney disease worldwide.

The various stages of the patient with CKD have an increased risk due to cardiovascular morbidity, premature mortality and / or decreased quality of life. Despite the increase in knowledge about possible mechanisms of the disease, the number of patients with CKD is increasing throughout the world, as shown by rises in deaths and the incidence and prevalence of patients with end-stage disease.

It is a consensus in medical science that CKD is a pathology, as a rule, asymptomatic until the final stage, for this reason it is compared to an iceberg since only the smallest part of it is visible, which makes it difficult to obtain accurate prevalence data Hill et al. (2016). In this way, the optimization of results related to prevention, diagnosis and treatment are essential to generate positive impacts for the patient and less costly for the health system.

The use of AI to help decision-making in health is of paramount importance, as experts estimate of 10% to 15% error rate in diagnoses. It also points out that medical misdiagnosis can be potentially reduced as knowledge and use of technologies increase Graber, Mark & Kissam (2012).

Gathering the elements of the application of AI in Health, there is a third element of paramount importance for optimizing decision-making in health, it is the application of the economics concepts in health, which seeks to maximize the results of health decisions considering aspects of improving the patient’s quality of life and cost of the procedure. The main economic evaluation methods used in the health area are: cost- benefit analysis, cost minimization analysis, cost-effectiveness analysis and cost-utility analysis.

The evolution of the knowledge frontier is carried out based on a systematic review of the main articles published on the research object. In this way, a survey was carried out on the use of machine learning models, in particular decision tree, whose application of cost-effectiveness models in the field of Health, notably nephrology.

The systematic review work becomes essential because it is a research methodology that uses scientific methods to assess the knowledge frontier and the contribution of relevant studies on the subject under study. As advocated by Moher et al. (2009) the systematic review uses statistical and qualitative methods to analyze and summarize the results of studies developed by science.

Thus, aiming to deepen studies on the subject, a systematic review was carried out with the main articles published in recent years in the main national and international journals, following the PRISMA methodology (Preferred Reporting Items for Systematic Reviews and Meta-Analyses):

  1. Formulate the research question;
  2. Carry out a systematic search;
  3. Select studies;
  4. Extract and Analyze the data;
  5. Synthesize the results;

2. Background

CKD is a progressive condition that affects kidney function and can lead to kidney failure and the need for treatment by dialysis or kidney transplantation. The prevalence of CKD has increased significantly in recent years, and hypertension and diabetes are the main risk factors for the disease.

According to a study published by Chen et al. (2019), the prevalence of CKD among adults with diabetes was 29.2% in the United States, and hypertension was identified as the main risk factor for disease progression.

The diagnosis of CKD is based on renal function tests, such as measuring the glomerular filtration rate (GFR) and urinalysis to assess the presence of proteins and red blood cells. A study published by Monet-Didailler et al. (2020) highlighted the importance of regular clinical assessment of patients with CKD to early identify disease-related complications and implement appropriate treatment measures.

CKD treatment includes measures to control risk factors, such as taking medication to control blood pressure and blood sugar, as well as lifestyle changes, such as weight loss, a balanced diet, and regular physical activity. Another article published by Wang et al. (2019), the combination of medications and lifestyle changes can be effective in controlling the progression of CKD.

Recent research has focused on developing new therapeutic approaches for CKD, including gene and immune therapies. A study published by Tian et al. (2019) described the successful use of a gene therapy to improve kidney function in mice with CKD. In addition, there is growing interest in primary prevention approaches for CKD, including awareness campaigns to identify risk factors early and implement measures to control them.

Using decision trees is an effective and affordable approach to machine learning. A decision tree is a functional representation that takes an array of attribute values as input and produces a single decision as output. Both input and output values can be discrete or continuous. This versatile structure allows the decision tree to deal with a wide range of data and generate relevant results for several problems Russel and Norvig (2010).

Decision tree models help clinicians make informed decisions about disease treatment and prevention, increasing prediction accuracy and improving the quality of patient care. In this sense, the decision tree can be combined with other machine learning techniques, such as neural networks and genetic algorithms, to further improve the accuracy of predictions.

The decision tree is widely used in conjunction with cost-effectiveness models in healthcare to assess the relationship between treatment costs and clinical outcomes. The goal is to determine the most effective and cost-effective strategy for treating a disease or medical condition.

As reported by Muennig (2016) who discusses the use of decision trees in conjunction with cost-effectiveness models in healthcare, mention is made of the flexibility and ability of decision tree models to integrate information from different sources. In this sense, it provides a detailed explanation of how to build decision tree models, including the steps of hypothesis definition, variable selection and sensitivity evaluation, it also presents applied examples of the use of the decision tree in conjunction with cost- effectiveness models, including the treatment of cardiovascular disease and the prevention of infections.

Considering Arlandis-Guzman (2011) who uses the decision tree together with cost- effectiveness models in drug analysis, he highlights the importance of the decision tree as a valuable tool to evaluate the effectiveness and profitability of different pharmacological treatment options.

As presented by Kuntz (2016) the use of the decision tree in conjunction with cost- effectiveness models in the analysis of effectiveness of health treatments, which highlights the advantages of using these models compared to other methodologies. It also discusses issues related to the challenges faced in modeling decision trees. The article presents applied examples of the use of the decision tree in conjunction with cost-effectiveness models, including the evaluation of different treatment options for chronic diseases.

By using a decision tree to model different treatment options for a disease, it is possible to evaluate the effectiveness of each treatment option based on cost and expected clinical outcomes, helping healthcare professionals and patients to make informed decisions, based in the best alternative.

3. Methodology

The systematic quest on the research problem “how to develop a cost-effectiveness analysis model using a decision tree to support clinical decision-making in nephrology”, was carried out following the search terms:

  • “cost-effectiveness analysis” OR “cost-benefit analysis” OR “cost-utility analysis” OR “decision trees” OR “Markov models” AND
  • “nephrology” OR “renal insufficiency” OR “chronic kidney disease” OR “dialysis” OR “kidney transplantation” AND
  • “modeling” OR “decision support techniques” OR “decision making” OR “decision theory” OR “decision analysis” AND
  • “clinical decision support systems” OR “medical informatics” OR “evidence- based medicine” OR “healthcare decision making” OR “clinical practice guidelines”

PubMed and Embase databases were used. The search results were as follows:

PubMed:

  • Results: 186
  • Titles and abstracts examined: 32
  • Articles included in the review: 9

Embase:

  • Results: 280
  • Titles and abstracts examined: 48
  • Articles included in the review: 9

From the analysis of titles and abstracts, studies that were not directly related to the objective of the review were excluded. The inclusion criteria for the final articles were:

  • Studies that presented cost-effectiveness analysis models using decision trees to support clinical decision-making in nephrology.
  • Studies that described model development in sufficient detail to allow replication.
  • Studies that addressed the use of evidence-based clinical decision support systems in nephrology.
  • Studies published in English.

Based on these criteria, 9 articles in PubMed and 9 articles in Embase were selected for inclusion in the systematic review.

The 9 selected articles in PubMed are:

  1. Surendra, Naren Kumar et al. Cost utility analysis of end stage renal disease treatment in Ministry of Health dialysis centres, Malaysia: Hemodialysis versus continuous ambulatory peritoneal dialysis. PloS one, v. 14, n. 10, p. e0218422, 2019.
  2. Evangelidis, Nicole et al. Lifestyle behaviour change for preventing the progression of chronic kidney disease: a systematic review. BMJ open, v. 9, n. 10, p. e031625, 2019.
  3. De Vries, Eline F.; RABELINK, Ton J.; VAN DEN HOUT, Wilbert B. Modelling the cost-effectiveness of delaying end-stage renal disease. Nephron, v. 133, n. 2, p. 89-97, 2016.
  4. Yaghoubifard, Safiye et al. Cost-effectiveness analysis of dialysis and kidney transplant in patients with renal impairment using disability adjusted life years in Iran. Medical journal of the Islamic Republic of Iran, v. 30, p. 390, 2016.
  5. Levin, A., Hemmelgarn, B., Culleton, B., Tobe, S., McFarlane, P., Ruzicka, M., … & Manns, B. (2008). Guidelines for the management of chronic kidney disease. CMAJ: Canadian Medical Association Journal, 179(11), 1154-1162.
  6. Yao, Jiaqi; JIANG, Xinchan; YOU, Joyce HS. Proactive therapeutic drug monitoring of adalimumab for pediatric Crohn’s disease patients: A cost‐ effectiveness analysis. Journal of Gastroenterology and Hepatology, v. 36, n. 9, p. 2397-2407, 2021.
  7. YUE, Suru et al. Machine learning for the prediction of acute kidney injury in patients with sepsis. Journal of translational medicine, v. 20, n. 1, p. 1-12, 2022.
  8. TISDALE, Rebecca L. et al. Cost-effectiveness of dapagliflozin for non-diabetic chronic kidney disease. Journal of general internal medicine, v. 37, n. 13, p. 3380-3387, 2022.
  9. ILYAS, Hamida et al. Chronic kidney disease diagnosis using decision tree algorithms. BMC nephrology, v. 22, n. 1, p. 1-11, 2021.

The 9 articles selected at Embase are:

  1. Aiyegbusi, O. L., Kyte, D., Cockwell, P., & Marshall, T. (2019). Development of a patient-centered renal replacement decision aid using conjoint analysis: an international survey of patient preferences for the relative importance of treatment outcomes in hemodialysis. BMC nephrology, 20(1), 1-10.
  2. Ghiasi, Attema, A. E., Steenhoek, A., & Oppe, M. (2017). Time trade-off utility weights for the EQ-5D-3L health states: Slovak Republic. Quality of Life Research, 26(6), 1587-1595.
  3. Wu, Xiaoxia et al. Effects of combined aerobic and resistance exercise on renal function in adult patients with chronic kidney disease: a systematic review and meta- analysis. Clinical Rehabilitation, v. 34, n. 7, p. 851-865, 2020.
  4. Rokhman, M. Rifqi et al. Economic evaluations of screening programs for chronic kidney disease: A systematic review. Value in Health, 2023.
  5. Zhao, Dan et al. Current progress in artificial intelligence-assisted medical image analysis for chronic kidney disease: A literature review. Computational and Structural Biotechnology Journal, 2023.
  6. Bowrin, Kevin et al. Cost-effectiveness analyses using real-world data: an overview of the literature. Journal of Medical Economics, v. 22, n. 6, p. 545-553, 2019.
  7. Bugeja, Ann et al. Kidney function, cardiovascular outcomes and survival of living kidney donors with hypertension: a systematic review protocol. BMJ open, v. 12, n. 12, p. e064132, 2022.
  8. Gasga, AE Frias et al. EE306 Cost-Effectiveness Analysis of Treatments in Chronic Kidney Disease in Mexico. Value in Health, v. 25, n. 7, p. S393-S394, 2022.
  9. Wang, Xue-Rong et al. Prevalence of coronary artery calcification and its association with mortality, cardiovascular events in patients with chronic kidney disease: a systematic review and meta-analysis. Renal Failure, v. 41, n. 1, p. 244-256, 2019.

4. Results and discussion

Aiming to present the analysis results and selection of the main articles on the subject, a table was created that shows the problem/objective of the research, the technique used, the results and the place of publication.

Table 1. Systematic review articles

ArticleProblem/ObjectiveMethodological techniqueResults
Cost utility analysis of end stage renal disease treatment in Ministry of Health dialysis centres, Malaysia: Hemodialysis versus continuous ambulatory peritoneal dialysis.To carry out a cost-effectiveness analysis between hemodialysis and peritoneal dialysis in the face of an exponential growth of dialysis patients in Malaysia.A Markov model was developed to investigate the cost- effectiveness of increasing incident CAPD adherence to 55% and 60% versus the current practice of 40% CAPD over a five-year time horizon.Outcome measures were cost per life year (LY), quality adjusted cost per LY (QALY), and incremental cost- effectiveness ratio (ICER) for the Markov model. Sensitivity analyzes were performed. Reduced CAPD use was associated with higher costs and a small depreciation in QALYs
Modelling the cost-effectiveness of delaying end-stage renal diseaseAs the incidence of end-stage renal disease (ESRD) is increasing, new therapies are being developed to delay ESRD. This study aims to build a generic model to estimate the cost- effectiveness of delaying ESRD in 7 European countries: the Netherlands, the United Kingdom, Germany, Italy, Spain, Finland and Hungary.Cost-effectiveness analysis was performed to estimate the potential impact of delaying ESRD in 7 European Union countries. The article investigates whether a new treatment is cost- effective, explicitly comparing its impact on costs and patient effectiveness in terms of QALYs. The relevant population for analysis consisted of patients with stage 4 CKD (CKD4), aged over 20 years, from the following European countries: Netherlands, United Kingdom, Germany, Italy, Spain, Finland and Hungary. The chosen countries represent a geographical distribution across Europe, with considerable differences  in transplantation rates and productivity values.A 1-year delay in ESRD resulted in an estimated gain of 0.6 QALYs and 0.3 years in productivity. Transplant access had a minimal impact, while       productivity savings had a significant impact. For a free 1-year deferral, social savings would vary between €8,000 in the UK and €17,000 in Germany. Applying thresholds of €20,000 to €40,000 per QALY gained, single cell- based therapy would be  economically acceptable       if it delayed end-stage renal disease by 0.2- 0.5 years. It would be a cost saving in a delay of more than 0.5 years. Continuous use of medicines is unlikely to be cost- effective for prices above 30,000 euros per year.
Cost- effectiveness analysis of dialysis and kidney transplant in patients with renal impairment using disability adjusted life years in Iran.Cross-sectional study was carried out to compare the cost-effectiveness of  three therapeutic methods of long- term hemodialysis, kidney transplantation from a living person and kidney transplantation from a cadaver using Disability Adjusted Life Years (DALY) using registry data of patients referred to Afzalipour Hospital in Kerman.It used data from the records of patients referred to Afzalipour Hospital in Kerman. Decision tree model and decision tree software (Tree Age pro 11) were used for data analysis.From the patient’s perspective, the CER of dialysis was 5.04 times greater than that of a living transplant and 6.15 times that of a cadaver donor transplant. From the hospital’s perspective, the average cost- effectiveness ratio of dialysis was 8.4 times greater than that of a living transplant and 14.07 times that of a cadaver       transplant. The lower the CE index, the greater the cost-benefit.
Proactive therapeutic drug monitoring of adalimumab for pediatric Crohn’s disease patients: A cost‐effectiveness analysis.Cost-effectiveness analysis of using therapeutic drug monitoring (TDM)to optimize tacrolimus dosing in kidney transplant patients in Singapore.The Markov model was constructed to estimate the outcomes of proactive versus reactive ADL TDM in a hypothetical cohort of pediatric patients with CD who were in remission on ADL maintenance treatment. Model data was derived from published      literature and public data. Model outputs included CD-related direct medical costs and quality-adjusted life years (QALYs). Sensitivity analyzes were performed to examine the robustness of the base case results.When compared with the reactive TDM group, the proactive TDM group saved 0.1960 QALYs at a lower cost of up to USD 2021 over a 3- year period in the base case analysis.    One-way sensitivity analysis showed that the     cost of ADL medications     is the most influential factor. Probabilistic sensitivity analysis of 10,000 Monte-Carlo simulations found that the proactive TDM group gained 0.1958 QALYs    and saved USD 2037.
Machine learning for the prediction of acute kidney injury in patients with sepsis.Establish and validate predictive models based on novel machine learning (ML) algorithms for AKI in critically ill patients with sepsis.Feature selection was performed using a Boruta algorithm. ML algorithms such as logistic regression (LR), k-nearest neighbors (KNN), support vector machine (SVM), decision tree, random forest, Extreme Gradient Boosting (XGBoost) and artificial neural network (ANN) were applied for model building using ten- fold cross-validation.A total of 3,176 critically ill patients with sepsis were included for analysis, of which 2,397 cases (75.5%) developed AKI during hospitalization. 36 variables were selected to build the model. The LR, KNN, SVM, decision tree, random forest, ANN, XGBoost, SOFA and SAPS II score models were established and obtained areas under the receiver Operating Characteristic curves of 0.7365, 0.6637, 0.7353, 0.7492, 0.7787, 0.7547, 0.821, 0.6457 and   0.7015, respectively. The XGBoost model had the best predictive performance in terms of discrimination, calibration and clinical application among all models.
Cost- effectiveness  of dapagliflozin for non-diabetic chronic kidney diseaseTo determine the cost-effectiveness of adding dapagliflozin to the standard treatment of non- diabetic CKD patients.Markov model with life time horizon and perspective of the US healthcare sector. Quality-adjusted life years (QALY), costs and incremental cost- effectiveness ratios (ICER), all discounted at 3% per year; total incidence of renal failure in renal replacement therapy; average years on renal replacement therapy. Dapagliflozin improved life expectancy and reduced CKD progression, the proportion of patients requiring renal replacement therapy, and the duration of renal replacement therapy in non- diabetic CKD patients. The use of dapagliflozin meets conventional cost- effectiveness criteria.
Chronic kidney disease diagnosis using decision tree algorithms.Proposes the application of decision trees in the development of clinical guidelines for the management of chronic kidney disease.Predict the various stages of CKD using machine learning classification algorithms on     the dataset obtained from the medical records of affected people. Specifically, we use Random Forest and J48 algorithms to obtain a sustainable and practicable model for detecting various stages of CKD with comprehensive medical accuracy.Comparative analysis of the results revealed that J48 predicted DRC at all stages better than random forest with an accuracy of 85.5%. The study also showed that J48 performs better compared to Random Forest.
Cost- Effectiveness Analysis of Treatments in Chronic Kidney Disease in Mexico.Carry out a complete economic evaluation of the type of cost- effectiveness analysis of treatment alternatives for chronic kidney disease in Mexico, from an institutional point of view.A Cost-Effectiveness Analysis (CEA) was performed as the primary analysis and a budget impact analysis as a secondary analysis in 2020. Resource utilization was determined for adult patients (18 years and older) with renal failure who are on some type of renal replacement therapy (RRT), whether peritoneal dialysis, hemodialysis, or kidney transplantation.The complete economic evaluation of cost-effectiveness in chronic renal failure establishes that for the Mexican population the dominant option for renal replacement therapy at age 6 is renal transplantation, followed by peritoneal dialysis and, finally, hemodialysis.

4. Conclusion

The bibliographic review, based on the parameters defined in the research methodology, allowed a broader and deeper view of the scientific evolution applied to the subject matter of this study.

The papers used in the systematic review present evidence that the cost-effectiveness model is indicated to optimize decision-making and the introduction of new technologies or medicines into the healthcare system. Markov models are also used as a methodology as a parameter for introduction into the healthcare system.

The use of a decision tree helps healthcare professionals understand the models, facilitating decision-making based on the support offered by artificial intelligence models.

From the reading of the published studies, it is noted that although well founded, there is still a shortage of published works, a fact that opens up the possibility of new applications and new frontiers of studies.

The use of artificial intelligence techniques in nephrology, notably using cost- effectiveness models, is subject to new applications, especially in the axis of tropical countries, such as in countries in South America and Africa.

The focus of the published studies comprises Europe, USA, Asia and the Middle East. Therefore, developing countries such as Brazil need to introduce guiding elements to support medical decision-making, as well as aiming to optimize the budget of the Unified Health System and gain quality of life for patients.

References

Aiyegbusi, O. L., Kyte, D., Cockwell, P., & Marshall, T. (2019). Development of a patient-centered renal replacement decision aid using conjoint analysis: an international survey of patient preferences for the relative importance of treatment outcomes in hemodialysis. BMC nephrology, 20(1), 1-10.

Arlandis-Guzman, Salvador et al. Cost-effectiveness analysis of antimuscarinics in the treatment of patients with overactive bladder in Spain: a decision-tree model. BMC urology, v. 11, p. 1-11, 2011.

Bowrin, Kevin et al. Cost-effectiveness analyses using real-world data: an overview of the literature. Journal of Medical Economics, v. 22, n. 6, p. 545-553, 2019.

Bugeja, Ann et al. Kidney function, cardiovascular outcomes and survival of living kidney donors with hypertension: a systematic review protocol. BMJ open, v. 12, n. 12, p. e064132, 2022.

Chen, Teresa K.; KNICELY, Daphne H.; GRAMS, Morgan E. Chronic kidney disease diagnosis and management: a review. Jama, v. 322, n. 13, p. 1294-1304, 2019.

De Vries, Eline F.; RABELINK, Ton J.; VAN DEN HOUT, Wilbert B. Modelling the cost-effectiveness of delaying end-stage renal disease. Nephron, v. 133, n. 2, p. 89-97, 2016.

Ene-Iordache, Bogdan et al. Chronic kidney disease and cardiovascular risk in six regions of the world (ISN-KDDC): a cross-sectional study. The Lancet Global Health, v. 4, n. 5, p. e307-e319, 2016.

Evangelidis, Nicole et al. Lifestyle behaviour change for preventing the progression of chronic kidney disease: a systematic review. BMJ open, v. 9, n. 10, p. e031625, 2019.

Gasga, AE Frias et al. EE306 Cost-Effectiveness Analysis of Treatments in Chronic Kidney Disease in Mexico. Value in Health, v. 25, n. 7, p. S393-S394, 2022.

Ghiasi, Bahareh et al. Quality of Life of patients with chronic kidney disease in Iran: Systematic Review and Meta-analysis. Indian journal of palliative care, v. 24, n. 1, p. 104, 2018.

Gomes, Cecilia Neta Alves Pegado. ONTODRC Rompendo o silêncio da Doença Renal Crônica. João Pessoa, 2018. Tese de Doutorado UFPB.

Graber, Mark L. et al. Cognitive interventions to reduce diagnostic error: a narrative review. BMJ quality & safety, v. 21, n. 7, p. 535-557, 2012.

Hill, Nathan R. et al. Global Prevalence of Chronic Kidney Disease – A Systematic Review and Meta-Analysis. 2016. PLoS ONE 11 (7): e0158765.

Ilyas, Hamida et al. Chronic kidney disease diagnosis using decision tree algorithms. BMC nephrology, v. 22, n. 1, p. 1-11, 2021.

Kuntz, Karen M. et al. Decision models in cost-effectiveness analysis. In: Cost- Effectiveness in Health and Medicine. Oxford University Press, New York, NY, 2016. p. 105-136.

Levin, Adeera et al. Guidelines for the management of chronic kidney disease. Cmaj, v. 179, n. 11, p. 1154-1162, 2008.

Levin, Adeera. The advantage of a uniform terminology and staging system for chronic kidney disease (CKD). Nephrology Dialysis Transplantation, v. 18, n. 8, p. 1446-1451, 2003.

Lunney, Meaghan. et al. Guidelines, policies, and barriers to kidney care: findings from a global survey. Kidney International Supplements. v.8, p. 30-40, 2018.

Moher, David et al. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. Annals of internal medicine, v. 151, n. 4, p. 264-269, 2009.

Muennig, Peter; BOUNTHAVONG, Mark. Cost-effectiveness analysis in health: a practical approach. John Wiley & Sons, 2016.

Rokhman, M. Rifqi et al. Economic evaluations of screening programs for chronic kidney disease: A systematic review. Value in Health, 2023.

Russell, Stuart J.; NORVIG, Peter. Artificial intelligence a modern approach. London, 2010.

Surendra, Naren Kumar et al. Cost utility analysis of end stage renal disease treatment in Ministry of Health dialysis centres, Malaysia: Hemodialysis versus continuous ambulatory peritoneal dialysis. PloS one, v. 14, n. 10, p. e0218422, 2019.

Tian, Xiaolin et al. The comparison of cognitive function and risk of dementia in CKD patients under peritoneal dialysis and hemodialysis: A PRISMA-compliant systematic review and meta-analysis. Medicine, v. 98, n. 6, 2019.

Tisdale, Rebecca L. et al. Cost-effectiveness of dapagliflozin for non-diabetic chronic kidney disease. Journal of general internal medicine, v. 37, n. 13, p. 3380-3387, 2022.

Wang, Xue-Rong et al. Prevalence of coronary artery calcification and its association with mortality, cardiovascular events in patients with chronic kidney disease: a systematic review and meta-analysis. Renal Failure, v. 41, n. 1, p. 244-256, 2019.

Monet-Didailler, Catherine et al. Outcome of children with Shiga toxin-associated haemolytic uraemic syndrome treated with eculizumab: a matched cohort study. Nephrology Dialysis Transplantation, v. 35, n. 12, p. 2147-2153, 2020.

Wu, Xiaoxia et al. Effects of combined aerobic and resistance exercise on renal function in adult patients with chronic kidney disease: a systematic review and meta-analysis. Clinical Rehabilitation, v. 34, n. 7, p. 851-865, 2020.

Yaghoubifard, Safiye et al. Cost-effectiveness analysis of dialysis and kidney transplant in patients with renal impairment using disability adjusted life years in Iran. Medical journal of the Islamic Republic of Iran, v. 30, p. 390, 2016.

Yao, Jiaqi; JIANG, Xinchan; YOU, Joyce HS. Proactive therapeutic drug monitoring of adalimumab for pediatric Crohn’s disease patients: A cost‐ effectiveness analysis. Journal of Gastroenterology and Hepatology, v. 36, n. 9, p. 2397-2407, 2021.

Yue, Suru et al. Machine learning for the prediction of acute kidney injury in patients with sepsis. Journal of translational medicine, v. 20, n. 1, p. 1-12, 2022.

Zhao, Dan et al. Current progress in artificial intelligence-assisted medical image analysis for chronic kidney disease: A literature review. Computational and Structural Biotechnology Journal, 2023.

1Centro de Informática – Universidade Federal da Paraíba (UFPB), antonioeconomia@gmail.com
2Centro de Informática – Universidade Federal da Paraíba (UFPB), natasha@ci.ufpb.br
³Centro de Informática – Universidade Federal da Paraíba (UFPB), leandro@ci.ufpb.br