BMC Medical Informatics and Decision Making最新文献

{"title":"Benchmarking bias in embeddings of healthcare AI models: using SD-WEAT for detection and measurement across sensitive populations.","authors":"Magnus Gray, Leihong Wu","doi":"10.1186/s12911-025-03102-8","DOIUrl":"10.1186/s12911-025-03102-8","url":null,"abstract":"","PeriodicalId":9340,"journal":{"name":"BMC Medical Informatics and Decision Making","volume":"25 1","pages":"258"},"PeriodicalIF":3.3,"publicationDate":"2025-07-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12247235/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144607371","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A holistic framework for intradialytic hypotension prediction using generative adversarial networks-based data balancing.","authors":"Hsuan-Ming Lin, JrJung Lyu","doi":"10.1186/s12911-025-03094-5","DOIUrl":"10.1186/s12911-025-03094-5","url":null,"abstract":"<p><strong>Background: </strong>Intradialytic Hypotension (IDH) is a frequent complication in hemodialysis, yet predictive modeling is challenged by class imbalance. Traditional oversampling methods often struggle with complex clinical data. This study evaluates an enhanced conditional Wasserstein Generative Adversarial Network with Gradient Penalty (CWGAN-GP) framework to improve IDH prediction by generating high-utility synthetic data for balancing.</p><p><strong>Methods: </strong>A CWGAN-GP was developed using multi-level hemodialysis data. Following rigorous preprocessing, including a strict temporal train-test split, the CWGAN-GP generated minority class samples exclusively on the training data. eXtreme Gradient Boosting (XGBoost) models were trained on the original imbalanced data and datasets balanced using the proposed CWGAN-GP method, benchmarked against traditional Synthetic Minority Over-sampling Technique(SMOTE) and Adaptive Synthetic Sampling Approach(ADASYN) balancing. Performance was evaluated using metrics sensitive to imbalance (e.g., Precision-Recall Area Under the Curve) and statistical comparisons, with SHapley Additive exPlanations (SHAP) analysis for interpretability.</p><p><strong>Results: </strong>The study population consisted of 40 chronic hemodialysis patients (45% male, mean age 66.30[Formula: see text] 10.68 years). An initial dataset, where intradialytic hypotension (IDH) events occurred in 14.85% of records (19,124 instances overall), was temporally split (75:25 ratio). This yielded an Original Training dataset of 95,856 samples (14.73% IDH rate) and a test set (15.21% IDH rate). From this Original Training dataset, a Generative Adversarial Network (GAN) was employed to construct a balanced dataset comprising 163,470 samples. The GAN Balanced dataset yielded the highest predictive performance, demonstrating statistically significant improvements over the Original Training dataset across metrics, including Precision-Recall Area Under the Curve (PR-AUC) (mean 0.735 vs 0.724) and Accuracy (mean 0.900 vs 0.892). In contrast, the GAN Augmented dataset (191,712 samples) showed mixed results (improved Accuracy/F1, decreased Receiver Operating Characteristic Curve Area Under Curve (ROC-AUC)/PR-AUC). In comparison, ADASYN (163,326 samples) and SMOTE (163,470 samples) balanced datasets significantly underperformed on PR-AUC. SHAP analysis identified Dialysis Date (as a proxy for temporal patterns like day-of-week) and hemodynamic indicators (e.g., Systolic Diastolic Difference, Previous Systolic Pressure) as key IDH predictors.</p><p><strong>Conclusion: </strong>The proposed CWGAN-GP framework effectively balances complex hemodialysis data, leading to significantly improved and interpretable IDH prediction models compared to standard approaches. This work supports leveraging advanced generative models like GAN to overcome data imbalance in clinical prediction tasks, which is pending further validation.</p>","PeriodicalId":9340,"journal":{"name":"BMC Medical Informatics and Decision Making","volume":"25 1","pages":"257"},"PeriodicalIF":3.3,"publicationDate":"2025-07-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12243283/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144599458","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Predicting antiretroviral therapy adherence status of adult HIV-positive patients using machine-learning Northwest, Ethiopia, 2025.","authors":"Kelemua Aschale Yeneakal, Gizaw Hailiye Teferi, Temesgen T Mihret, Abraham Keffale Mengistu, Sefefe Birhanu Tizie, Maru Meseret Tadele","doi":"10.1186/s12911-025-03106-4","DOIUrl":"10.1186/s12911-025-03106-4","url":null,"abstract":"<p><strong>Background: </strong>Adherence with Anti-Retroviral Therapy (ART) reduces viral load, as well as HIV-related morbidity and mortality. Despite the expanded availability of ART, non-adherence remains a series problem, leads increased viral load, a decline CD4 cell count, and the development of drug resistance. HIV care is currently showing promise with the use of machine learning algorithms for early prediction of future non-adherence. However, as to researcher's Knowledge, there was limited research supporting this evidence in the country. Therefore, the primary aim of this study was to predict ART adherence status using machine learning models and to identify the most important predictors of Adherence at Debre Markos comprehensive specialized hospital.</p><p><strong>Methods: </strong>Secondary data was collected from ART database of Debre Markos comprehensive specialized hospital, spanning from 2005 to 2024. The dataset was split into training (80%) and testing (20%) sets. To address class imbalance, the Synthetic Minority Oversampling Technique (SMOTE) was applied to the training data. Seven machine learning algorithms: support vector machine, random forest, decision tree, logistic regression, gradient boosting, K-nearest neighbors, and artificial neural network were trained. The model performance was evaluated using ROC-AUC, F1 score, accuracy, precision, and recall. To identify important predictor we employed feature importance technique.</p><p><strong>Result: </strong>Out of 4640 patients, who were on antiretroviral therapy, 63.56% (n = 2949) were females, with mean age of 41.8 years (SD ± 11.50). The majority age group was between 40 and 59 years (n = 2152) 46.38% and 98.1% of patients had good adherence while 1.9% had poor adherence. Among the machine learning models tested, the gradient boosting algorithm performed better than all other algorithms with (Accuracy = 0.78, Sensitivity = 0.76, F1score = 0.78, AUC = 0.76). Age, regimen, WHO clinical stage, nutritional status, address status, sex, weight, recent CD4 cell count, viral load and ART dose per day were identified as the most important predictors for adherence status.</p><p><strong>Conclusion: </strong>The study developed a gradient boosting model for predicting adherence status. Age, regimen, WHO clinical stage, nutritional status, address status, sex, weight, recent CD4 cell count, viral load and ART dose per day were the most important predictors for adherence status.</p><p><strong>Clinical trial number: </strong>Not applicable.</p>","PeriodicalId":9340,"journal":{"name":"BMC Medical Informatics and Decision Making","volume":"25 1","pages":"259"},"PeriodicalIF":3.3,"publicationDate":"2025-07-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12247312/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144607372","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Coding mechanisms for main condition in ICD-11.","authors":"Hude Quan, Olafr Steinum, Danielle A Southern, William A Ghali","doi":"10.1186/s12911-025-03069-6","DOIUrl":"10.1186/s12911-025-03069-6","url":null,"abstract":"<p><p>Countries have been routinely abstracting health data from hospital charts and coding conditions using ICD-10. A main condition must be assigned to each admission. However, the definition of main condition is inconsistent across countries, and may be based on (1) the initial reason for admission; (2) the reason for admission, as understood at the end of the hospital stay; and (3) the condition that consumed the most hospital resources or hospital days. Now, ICD-11 standardizes the coding schema for main condition. This paper describes the ICD-11 coding guidelines for main condition and discusses their implications for data comparability.</p>","PeriodicalId":9340,"journal":{"name":"BMC Medical Informatics and Decision Making","volume":"21 Suppl 6","pages":"387"},"PeriodicalIF":3.3,"publicationDate":"2025-07-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12243148/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144607373","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Development of an explainable machine learning model for predicting device-related pressure injuries in clinical settings.","authors":"Yijie Qian, Hongying Pan, Jun Chen, Hongyang Hu, Mei Fang, Chen Huang, Yihong Xu, Yang Gao","doi":"10.1186/s12911-025-03090-9","DOIUrl":"10.1186/s12911-025-03090-9","url":null,"abstract":"<p><strong>Background: </strong>Device-related pressure injury (DRPI) is a prevalent and severe problem for patients using medical devices. Timely identification of patients at high risk of DRPI is crucial for healthcare providers to make informed decisions and prevent DRPI quickly. Given the rapid advancements in computer technology, we aimed to develop an interpretable artificial intelligence (AI) model for predicting DRPI, utilizing SHAP (SHapley Additive exPlanations) to enhance the model's transparency and provide insights into feature importance.</p><p><strong>Methods: </strong>We enrolled 675 study subjects (225 in the DRPI group and 450 in the non-DRPI group) from a single medical center between January 2019 and December 2020. Python was used to perform classification models, including extreme gradient boosting (XGBoost), random forest (RF), decision tree (DT), Logistic Regression (LR), support vector machine (SVM), and K-Nearest neighbors (KNN). We evaluated the performance of the six models using area under the ROC curve (AUC), specificity, accuracy, and sensitivity, with the dataset split into a 80% training set and a 20% testing set. We utilized several analyses, such as SHAP and Uniform Manifold Approximation and Projection (UMAP), to explore the potential contribution of different characteristics in our risk prediction models.</p><p><strong>Results: </strong>In the test set, XGBoost model outperformed the other models (AUC = 0.964). The interpretation of the model using SHAPscores revealed that the length of stay, instrument type, emergency admissions, instrument material, and instrument duration of use are the top five most important features in predicting DRPI.</p><p><strong>Conclusion: </strong>Our study demonstrated that the development of DRPI in patients can be accurately predicted using the machine learning (ML) model. The findings not only provide clinical caregivers with a valuable framework to identify patients at high risk of DRPI, but also lay the groundwork for developing targeted preventive strategies and personalized interventions.</p>","PeriodicalId":9340,"journal":{"name":"BMC Medical Informatics and Decision Making","volume":"25 1","pages":"256"},"PeriodicalIF":3.3,"publicationDate":"2025-07-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12239401/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144599459","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Streamlining medical software development with CARE lifecycle and CARE agent: an AI-driven technology readiness level assessment tool.","authors":"Steven N Hart, Patrick L Day, Christopher A Garcia","doi":"10.1186/s12911-025-03099-0","DOIUrl":"10.1186/s12911-025-03099-0","url":null,"abstract":"<p><strong>Background: </strong>Developing medical software requires navigating complex regulatory, ethical, and operational challenges. A comprehensive framework that supports both technical maturity and clinical safety is essential for effective artificial intelligence and machine learning system deployment. This paper introduces the Clinical Artificial Intelligence Readiness Evaluator Lifecycle and the Clinical Artificial Intelligence Readiness Evaluator Agent-a framework and AI-driven tool designed to streamline technology readiness level assessments in medical software development.</p><p><strong>Methods: </strong>We developed the framework using an iterative process grounded in collaborative stakeholder analysis. Key institutional stakeholders-including clinical informatics experts, data engineers, ethicists, and operational leaders-were engaged to identify and prioritize the regulatory, ethical, and technical requirements unique to clinical AI/ML development. This approach, combined with a thorough review of existing methodologies, informed the creation of a lifecycle model that guides technology maturation from initial concept to full deployment. The AI-driven tool was implemented using a retrieval-augmented generation strategy and evaluated through a synthetic use case (the Diabetes Outcome Predictor). Evaluation metrics included the proportion of correctly addressed assessment questions and the overall time required for automated review, with human adjudication validating the tool's performance.</p><p><strong>Results: </strong>The findings indicate that the proposed framework effectively captures the complexities of clinical AI development. In the synthetic use case, the AI-driven tool identified that 32.8% of the assessment questions remained unanswered, while human adjudication confirmed discrepancies in 19.4% of these instances. These outcomes suggest that, when fully refined, the automated assessment process can reduce the need for extensive multi-stakeholder involvement, accelerate project timelines, and enhance resource efficiency.</p><p><strong>Conclusions: </strong>The Clinical Artificial Intelligence Readiness Evaluator Lifecycle and Agent offer a robust and methodologically sound approach for evaluating the maturity of medical AI systems. By integrating stakeholder-driven insights with an AI-based assessment process, this framework lays the groundwork for more streamlined, secure, and effective clinical AI development. Future work will focus on optimizing retrieval strategies and expanding validation across diverse clinical applications.</p>","PeriodicalId":9340,"journal":{"name":"BMC Medical Informatics and Decision Making","volume":"25 1","pages":"254"},"PeriodicalIF":3.3,"publicationDate":"2025-07-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12236009/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144590532","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A novel method to predict the haemoglobin concentration after kidney transplantation based on machine learning: prediction model establishment and method optimization.","authors":"Songping He, Xiangxi Li, Fangyu Peng, Jiazhi Liao, Xia Lu, Hui Guo, Xin Tan, Yanyan Chen","doi":"10.1186/s12911-025-03060-1","DOIUrl":"10.1186/s12911-025-03060-1","url":null,"abstract":"<p><strong>Background: </strong>Anaemia is a common complication after kidney transplantation, and the haemoglobin concentration is one of the main criteria for identifying anaemia. Moreover, artificial intelligence methods have developed rapidly in recent years, are widely used in the medical field and have achieved good results.</p><p><strong>Objective: </strong>To optimize the process of constructing a clinical prediction model based on machine learning and improve related technologies. A classification prediction model for the haemoglobin concentration after kidney transplantation was constructed.</p><p><strong>Methods: </strong>Real-world data from 854 kidney transplant patients in a Grade A tertiary hospital were retrospectively extracted. An imputation method combining the K-nearest neighbour algorithm and multilayer perceptron was used to fill in missing values in the dataset. Recursive feature elimination and extreme gradient boosting were used to rank and screen the importance of patient features and reduce the dimensionality of the features. Before the classification prediction model was established, the number of classification categories was determined first, and the optimal ideal cluster was approximated by the ideal cluster under each classification number and the similarity between the ideal cluster and the actual cluster. Finally, five kinds of machine learning methods, random forest, extreme gradient boosting, light gradient boosting machine, linear support vector classifier and support vector machine, were used to establish classification prediction models, and error-correcting output codes were used to optimize each model. A classification prediction model for abnormal haemoglobin concentrations after kidney transplantation was constructed, and the prediction effect was verified.</p><p><strong>Results: </strong>The imputation method combining the K-nearest neighbour algorithm and multilayer perceptron has a better effect on the imputation of missing values than do the commonly used imputation methods. Among the machine learning methods used for modelling, the prediction results of the tree model are improved to a certain degree after the error-correcting output code optimization. The final model with the best effect is optimized extreme gradient boosting, and the prediction accuracies before and after model optimization are 85.98% and 87.22%, respectively.</p><p><strong>Conclusions: </strong>The accuracy of the machine learning classification prediction model established by the optimized modelling method and process reached 87.22%, which can assist doctors in preoperative risk prediction.</p>","PeriodicalId":9340,"journal":{"name":"BMC Medical Informatics and Decision Making","volume":"25 1","pages":"255"},"PeriodicalIF":3.3,"publicationDate":"2025-07-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12236034/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144590531","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Association between blood urea nitrogen to albumin ratio and 28-day mortality in ICU patients with acute respiratory failure: a retrospective analysis of MIMIC-IV database.","authors":"Zhen Li, Jun Xie, Qian He, Chong Li","doi":"10.1186/s12911-025-03100-w","DOIUrl":"10.1186/s12911-025-03100-w","url":null,"abstract":"<p><strong>Purpose: </strong>The effect of the blood urea nitrogen-to-albumin ratio (BAR) on 28-day mortality in intensive care unit (ICU) patients with acute respiratory failure (ARF) is unknown.</p><p><strong>Methods: </strong>Patients diagnosed with ARF were screened and randomly divided into training and validation sets (7:3) on the basis of the ICD-9 and ICD-10 diagnosis codes in the Medical Information Mart for Intensive Care IV (v.2.2) database. The primary outcome was the 28-day mortality after ICU admission. The training set was categorized into the low- and high-BAR groups on the basis of the optimal BAR cutoff values for 28-day mortality determined via receiver operating characteristic analysis. The clinical significance of the BAR was evaluated by the areas under the curve (AUCs), decision curve analysis (DCA), Kaplan-Meier (K-M) survival curve, logistic regression analyses and subgroup analysis.</p><p><strong>Results: </strong>In total, 2,766 patients were included. The 28-day mortality rate was 30.2%. The AUCs and 95% confidence interval (CI) for the BAR were AUC 0.644 (95%CI, 0.618 to 0.671) in training set. Multivariate logistic regression revealed that the BAR was an independent factor affecting the prognosis of ARF in both training and validation sets. K-M curves revealed a significant difference in 28-day mortality between the low- and high-BAR groups (p < 0.001). DCA showed moderate performance. No obvious interaction was found by subgroup analysis in most subgroups.</p><p><strong>Conclusion: </strong>The present work revealed that elevated BAR was significantly associated with worse 28-day mortality in patients with any cause of ARF. It remains to be shown whether retrospective analysis of an independent cohort can confirm the high predictive value of BAR.</p>","PeriodicalId":9340,"journal":{"name":"BMC Medical Informatics and Decision Making","volume":"25 1","pages":"253"},"PeriodicalIF":3.3,"publicationDate":"2025-07-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12232855/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144574865","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"MIDAS: a technology-enabled hub-and-spoke system for the collection and dissemination of high-quality medical datasets in India.","authors":"Dibyajyoti Maity, Rohit Satish, Raghu Dharmaraju, Vijay Chandru, Rajesh Sundaresan, Harpreet Singh, Debnath Pal","doi":"10.1186/s12911-025-03092-7","DOIUrl":"10.1186/s12911-025-03092-7","url":null,"abstract":"<p><strong>Background: </strong>The need for better AI models fuels the demand for larger and larger high-quality datasets with significant diversity. Over the years, many medical imaging datasets have been published globally, but existing datasets do not contain enough samples from the population of the Indian subcontinent, leading to subpar performance of developed AI models when deployed in India. The Medical Imaging and Information Datasets (MIDAS) India initiative was launched to address this by developing standards, protocols, and policies for gathering medical imaging data nationwide.</p><p><strong>Methods: </strong>MIDAS employs a hub-and-spoke system for data collection, where each thematic hub works with a set of spokes to collect data for a specific disease or medical condition from primary, secondary, and tertiary health centers. The data gathering is guided by standard operating procedures developed from the collaborative efforts of the participating medical institutions. The annotation protocols are based on a combination of gold-standard tests and/or agreement between experts to achieve the required labeling accuracy, depending on the data type and the intended purpose of the dataset.</p><p><strong>Results: </strong>The MIDAS platform is accessible at https://midas.iisc.ac.in/ . Two datasets are already available on MIDAS, one for oral cancer and another for dural-based pathologies, for free download. Many others are under development and review. Annotated and curated data are also available under various licenses as shared by the platform partners for the registered users. The datasets use standardized ontologies for annotations at both image and pixel-level regions of interest. The annotations undergo a review process before being published and accessible for download. Standards and guidelines for creating the datasets are evolving due to the complexity of the elements involved. Challenges are steeper, especially for data originating from early or pre-onset stages of diseases, such as dysplasia in oral cancer, where the manifestation of the disease feature(s) is sometimes unclear.</p><p><strong>Conclusion: </strong>MIDAS India aims to catalyze the AI-driven transformation of healthcare by providing high-quality annotated imaging data tailored to local needs. It supports innovation, regulatory assessment, and clinical adoption of AI tools, serving as a scalable model for other countries looking to build similar data infrastructure to enhance digital healthcare delivery.</p>","PeriodicalId":9340,"journal":{"name":"BMC Medical Informatics and Decision Making","volume":"25 1","pages":"252"},"PeriodicalIF":3.3,"publicationDate":"2025-07-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12232576/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144574869","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"medAL-suite: A software solution for creating and deploying complex clinical decision support algorithms.","authors":"Ludovico Gennaro Cobuccio, Vincent Faivre, Rainer Tan, Alan Vonlanthen, Fenella Beynon, Emmanuel Barchichat, Alain Fresco, Quentin Girard, Sinan Ucak, Sylvain Schaufelberger, Ibrahim Evans Mtebene, Peter Agrea, Emmanuel Kalisa, Gillian A Levine, Martin Norris, Sabine Renggli, Alix Miauton, Lisa Cleveley, Kristina Keitel, Julien Thabard, Valérie D'Acremont, Alexandra V Kulinkina","doi":"10.1186/s12911-025-03077-6","DOIUrl":"10.1186/s12911-025-03077-6","url":null,"abstract":"<p><strong>Background: </strong>Sub-optimal healthcare quality in low-resource settings is attributed in part to poor adherence to clinical guidelines. Clinical decision support systems (CDSS) help to integrate guideline-based algorithms into logical workflows and improve adherence to evidence-based recommendations, and hence quality of care. However, the process of translating paper-based guidelines into electronic algorithmic formats is often complex, inefficient, expensive, and error-prone due to reliance on advanced software development skills and clinical knowledge.</p><p><strong>Methods: </strong>In response to these challenges, we developed open-source software called the Medical Algorithm Suite (medAL-suite), consisting of four components, with a primary goal of increasing efficiency, accuracy, and transparency of CDSS creation by giving experienced clinicians, rather than software developers, greater control over the process. At the heart of the software suite is the medAL-creator that allows clinicians to design algorithms using a code-free drag-and-drop interface. Algorithms are subsequently automatically deployed in medAL-reader to service level clinicians in health facilities. CDSS implementers use medAL-data and medAL-hub to manage configuration, versioning, and deployment.</p><p><strong>Results: </strong>Since its development, the medAL-suite has been used to digitalize complex primary care guidelines and deployed in large-scale clinical studies in Tanzania, Rwanda, Kenya, Senegal, and India, leading to notable outcomes such as the reduction of inappropriate antibiotic prescriptions and improvement in care quality. Over 300,000 pediatric outpatient consultations have been completed in Rwanda and Tanzania to date using the digital algorithm.</p><p><strong>Discussion: </strong>The medAL-suite focused on democratized development, process-centric design, point-of-care utility, touch-screen interface, low cost, and low power consumption to contribute to sustainable digital systems in low-resource settings. Important future developments and adaptations as the software evolves should emphasize interoperability and scalability, primarily via integrating CDSS functionality into electronic medical records for a streamlined user experience that supports improved service quality at the point-of-care.</p><p><strong>Clinical trial number: </strong>Not applicable.</p>","PeriodicalId":9340,"journal":{"name":"BMC Medical Informatics and Decision Making","volume":"25 1","pages":"249"},"PeriodicalIF":3.3,"publicationDate":"2025-07-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12226914/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144564612","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}