Solmaz Sohrabei

The convergence of artificial intelligence (AI) and wearable monitoring devices is revolutionizing the management of high-risk pregnancies, offering unprecedented opportunities for early detection, personalized care, and improved outcomes. These technologies address critical challenges in maternal and fetal health, especially in resource-limited settings, by enabling proactive and continuous care. This systematic review specifically focuses on the integration of artificial intelligence and wearable devices for managing high-risk pregnancies and highlights regulatory and ethical challenges. Databases like PubMed (Medline), Scopus, and Web of Science were explored for pertinent studies (from 2015 to 2024). Articles were chosen based on specific inclusion and exclusion criteria, and essential data including technology type, AI algorithms employed, and their influence on perinatal care were collected. The Risk of Bias in the studies was evaluated utilizing standard assessment tools such as PROBAST+AI, and study quality was assessed with TRIPOD+AI. Ultimately, the findings were examined through narrative analysis, leading to the identification of challenges and future research directions. This systematic review examined 16 studies focusing on the application of AI and digital innovations in maternal and fetal healthcare. The studies investigated various AI techniques, including machine-learning models for predicting risks to maternal health, deep learning for monitoring fetal well-being, and wearable sensor technologies for remote health tracking. Many studies produced encouraging outcomes, with AI models reaching high levels of accuracy in early detection, risk assessment, and real-time patient monitoring. However, issues such as dataset biases, generalizability, and ethical concerns continue to pose significant challenges. Additional research is necessary to validate these AI-driven methods across diverse populations to ensure equitable and effective healthcare outcomes. The integration of technologies has the potential to significantly improve the management of high-risk pregnancies. However, a critical regulatory gap exists that must be addressed to ensure that these technologies are used ethically, safely, and effectively. This will involve the development of new regulatory frameworks, a focus on transparency and accountability, and a commitment to addressing bias and data security to provide the best care possible. AI has the potential to transform maternity services by improving risk prediction, enabling remote monitoring, and supporting clinical decisions. However, realizing this potential requires addressing limitations related to data quality, algorithm accuracy, user privacy, and ethical considerations.

Maternal and newborn mortality remain stubbornly high in low-resource settings, driven by limited access to timely, personalized and emotionally supportive care during pregnancy. Large language models/generative pre-trained transformer (LLM/GPTs), particularly GPT-driven conversational agents, have emerged as scalable, versatile digital health tools capable of delivering evidence-based information, mental health support and risk stratification for complications such as preeclampsia, gestational diabetes and preterm birth. This systematic review aimed to synthesize global evidence on the design, implementation and effectiveness of LLM/GPT-powered chatbots for precision maternal and newborn health engagement. Following PRISMA 2020 guidelines, we searched MEDLINE, Embase, CINAHL, Web of Science, Inspec and IEEE Xplore from January 2015 to November 2025. We included 15 peer-reviewed studies (published 2021–2025) that developed or evaluated GPT-based or equivalent generative conversational agents for pregnant individuals or their partners. Quality appraisal used EQUATOR tools and artificial intelligence-specific frameworks; two reviewers independently assessed risk of bias. The 15 studies (total participants >12 000) covered 12 countries. LLM/GPT-powered chatbots outperformed LLM/GPT systems in conversational naturalness, topic diversity and user satisfaction (mean acceptability scores 85–94%). Core functions included real-time psychoeducation (n = 14 studies), mental health screening and behavioral activation (n = 11/15), partner engagement (n = 6/15), and predictive risk modeling for adverse outcomes (n = 9). LLM models achieved high diagnostic concordance with clinicians for gestational diabetes (area under the curve 0.88 to 0.94) and preeclampsia warning signs. User retention ranged from 62 to 78% over 6 months, with the strongest engagement among prim parous and underserved populations. No serious harms were reported. LLM/GPT-driven conversational agents represent a breakthrough in accessible, personalized maternal and newborn health support. They effectively bridge gaps in high-risk pregnancy, emotional care, risk detection and partner involvement while maintaining safety and cultural adaptability. These findings strongly support rapid integration of LLM/GPT-powered chatbots into routine antenatal care pathways, particularly in low- and middle-income settings. The study was registered with PROSPERO (CRD420251230253).

Objectives: In physical reality, mobility limitations often hinder the ability of older individuals to fully interact with the world. Aging increases the likelihood of a decline in social activities. New technologies, including immersion technologies, are increasingly gaining attention as society grapples with the challenges of an aging population. The metaverse offers a compelling opportunity to reimagine the perspective of empowering older individuals.Methods: A systematic review was conducted on the PubMed, IEEE, Scopus, and Web of Science databases in December 2024 using a search strategy composed of the keywords “elderly,” “metaverse,” “augmented reality” and “virtual reality,” along with their synonyms in the titles of the articles. Descriptive, qualitative, review, and non‐English studies were excluded. The quality assessment of the articles and their bias evaluation were performed based on the JBI index and PRISMA guidelines.Results: The search yielded 1153 articles, 37 of which met the study entry conditions. Thirty studies utilized virtual reality technology for the rehabilitation of elderly people. Most studies aimed to improve symptoms of Alzheimer's disease, physical activity, and cognitive impairment. Seven studies focused on the design and use of immersive technologies to improve Parkinson's symptoms and motor impairments resulting from strokes. Four studies focused on designing environments and games that could help elderly people engage in physical activity and exercise using immersive technologies. The quality of life of elderly people improved in line with the use of these technologies, and symptom weakness improved.Conclusions: The metaverse can serve as a comprehensive and supportive platform for elderly people. The development of various digital applications or applications covering health, banking, financial affairs, shopping, and social media for performing daily activities from home via an online mode can assist elderly people. The metaverse offers innovative solutions to inspire and support elderly people in staying active. VR and AR technologies provide a dynamic platform for participation in exercise programs, recreational activities, and personalized health programs for elderly people.

Background: Breast cancer diagnostic data is complex and accompanied by noise. Quantum machine learning can enhance the accuracy, efficiency, and scalability of artificial intelligence algorithms and has applications in various fields such as drug discovery and personalized medicine. Methods: In the systematic review conducted, the databases PubMed, Embase, Scopus, and Web of Science were searched in December 2024. The search strategy included the keywords "Breast Cancer," "Artificial Intelligence," and "Quantum machine learning" along with their synonyms in article titles. Descriptive, qualitative, review, and non-English studies were excluded. The qualitative evaluation of the articles and the assessment of their bias were determined based on the Joanna Briggs Institute (JBI) indicators checklist. Results: Twenty-nine studies utilizing artificial intelligence models for personalized breast cancer management were selected. Seventeen studies employing various deep learning methods achieved satisfactory results in predicting treatment response and prognosis, effectively contributing to the personalized management of breast cancer. Twenty-six studies demonstrated that machine learning methods could enhance the processes of classification, screening, diagnosis, and prognosis of breast cancer. The methods most frequently used in modeling were quantum support vector machine (QSVM), quantum convolutional neural network (QCNN), and quantum neural network (QNN), with an average AUC of 0.91. Additionally, the average accuracy, sensitivity, specificity, and precision indices of the models ranged from 90% to 96%. Conclusion: Quantum computing can address some challenges arising from the increasing complexity and size of artificial intelligence models. Overall, the combination of artificial intelligence and quantum computing can significantly accelerate the drug discovery process and the development of effective drugs.

Background and objective: Breast cancer has become a common cancer in women. The early diagnosis of breast cancer has beneficial effects on the life patients. Due to difficulties in the disease, data mining techniques could help to facilitate the diagnosis, the current study amid to compare the efficiency of Genetic Algorithm (GA) and Multi- Layer Perceptron (MLP) in the diagnosis of the breast cancer. Methods: The database used in this paper is provided by Tehran university of Motamed Cancer Institute (MCI) breast cancer research center. This database included 7,625 records; there were 4,008 patients (52.4%) with breast cancers (malignant) and the remaining 3,617 patients (47.6%) without breast cancers (benign). GA and MLP models were developed using 14 fields (risk factor) of the database. The present study divided the data into 10 folds where 1 fold for testing and 9 folds for training as a way of validating the 10-fold crossover validation. Ultimately, the comparison of the models was made based on sensitivity, specificity, accuracy and ROC indicators. Findings: Sensitivity, specificity, accuracy and ROC under curve of the MLP model were 0.815, 76.27, 79.71 and 81.24 respectively. For the GA model, the note indicators respectively reported: 0.884, 86.32, 87.67 and 88.50. There was statistical significant difference between indicators of the two models (p-value<0.0001). Conclusion: Both models had acceptable efficiencies in diagnosing breast cancer, that GA had better efficiency. The number of breast cancer risk factors and number of database records can cause different sensitivity, specificity, accuracy and ROC indicators. More breast cancer risk factors such as mutation types could help to developing more efficient GA and ANN models. Keywords Diagnosis model; Breast cancer; MLP; GA

Background and Aims: Infection with Covid‐19 disease can lead to mortality in a short time. Early prediction of the mortality during an epidemic disease can save patients' lives through taking timely and necessary care interventions. Therefore, predicting the mortality of patients with Covid‐19 using machine learning techniques can be effective in reducing mortality rate in Covid‐19. The aim of this study is to compare four machine‐learning algorithm for predicting mortality in Covid‐19 disease.Methods: The data of this study were collected from hospitalized patients with COVID‐19 in five hospitals settings in Tehran (Iran). Database contained 4120 records, about 25% of which belonged to patients who died due to Covid‐19. Each record contained 38 variables. Four machine‐learning techniques, including random forest (RF), regression logistic (RL), gradient boosting tree (GBT), and support vector machine (SVM) were used in modeling.Results: GBT model presented higher performance compared to other models (accuracy 70%, sensitivity 77%, specificity 69%, and the ROC area under the curve 0.857). RF, RL, and SVM models with the ROC area under curve 0.836, 0.818, and 0.794 were in the second and third places.Conclusion: Considering the combination of multiple influential factors affecting death Covid‐19 can help in early prediction and providing a better care plan. In addition, using different modeling on data can be useful for physician in providing appropriate care.

Introduction. In recent years, the use of dashboards in healthcare has been considered an effective approach for the visual presentation of information to support clinical and administrative decisions. Effective and efficient use of dashboards in clinical and managerial processes requires a framework for the design and development of tools based on usability principles. Objectives. The present study is aimed at investigating the existing questionnaires used for the usability evaluation framework of dashboards and at presenting more specific usability criteria for evaluating dashboards. Methods. This systematic review was conducted using PubMed, Web of Science, and Scopus, without any time restrictions. The final search of articles was performed on September 2, 2022. Data collection was performed using a data extraction form, and the content of selected studies was analyzed based on the dashboard usability criteria. Results. After reviewing the full text of relevant articles, a total of 29 studies were selected according to the inclusion criteria. Regarding the questionnaires used in the selected studies, researcher‐made questionnaires were used in five studies, while 25 studies applied previously used questionnaires. The most widely used questionnaires were the System Usability Scale (SUS), Technology Acceptance Model (TAM), Situation Awareness Rating Technique (SART), Questionnaire for User Interaction Satisfaction (QUIS), Unified Theory of Acceptance and Use of Technology (UTAUT), and Health Information Technology Usability Evaluation Scale (Health‐ITUES), respectively. Finally, dashboard evaluation criteria, including usefulness, operability, learnability, ease of use, suitability for tasks, improvement of situational awareness, satisfaction, user interface, content, and system capabilities, were suggested. Conclusion. General questionnaires that were not specifically designed for dashboard evaluation were mainly used in reviewed studies. The current study suggested specific criteria for measuring the usability of dashboards. When selecting the usability evaluation criteria for dashboards, it is important to pay attention to the evaluation objectives, dashboard features and capabilities, and context of use.