Network Analysis-Driven Machine Learning Model for Identifying High-Cost Stroke Inpatients Using Hospital Discharge Data: Retrospective Study.

Cancer affects not only physical health but also the emotional, social, and psychological well-being of patients and their close networks across the cancer trajectory []. Cancer survivors often face symptoms, lifestyle adjustments, and report information needs that persist long after cancer treatment []. Although cancer survivors value continued personal follow-up care [,], it is rarely achievable in routine practice due to limited health care resources, competing priorities, geographic barriers, and the growing number of individuals living with and beyond cancer []. These challenges highlight the need for scalable solutions that can complement traditional survivorship care. Patient-facing digital health technologies (DHTs), such as mobile apps and digital therapeutics, may help bridge these gaps by delivering accessible, personalized, and timely support to patients in diverse settings []. Unlike DHTs designed for health care professionals, patient-facing DHTs are directly operated by patients themselves, enabling self-monitoring, symptom reporting, psychoeducation, and behavioral interventions in daily life []. Incorporating AI components into DHTs might further enhance their potential impact [,]. For instance, machine learning (ML) and natural language processing (NLP) can facilitate the processing and interpretation of large volumes of heterogeneous data, including symptom reports, sensor data, and medical records []. These technologies can thereby enable patient-facing DHTs to provide tailored feedback, adapt digital interventions to individual needs, or identify moments when users are most receptive to receiving an intervention [,]. However, these potential benefits should be weighed against important concerns. Embedding AI into patient-facing DHTs may introduce additional risks to data privacy and exacerbate existing gaps in digital access [], and consequently in access to health care. Furthermore, understanding how patients perceive and use these tools is necessary, as their experience will strongly influence engagement and uptake []. Although interest in patient-facing DHTs is increasing, few trials examine their effects from the patients’ perspective []. Patient-reported outcomes (PROs) are critical indicators of how individuals perceive their health, functioning, and quality of life, as well as the impact of interventions on their daily lives []. PROs capture dimensions of health that cannot be observed or measured by clinicians alone and are particularly relevant in chronic and oncological conditions where maintaining quality of life is a primary treatment goal []. Health-related quality of life (HRQOL) represents a key multidimensional construct spanning physical, psychological, and social domains [,]. Various instruments exist to assess HRQOL, ranging from generic measures to cancer-specific scales, each assessing different facets of survivors’ lived experiences [,]. However, it remains unclear whether AI-enabled DHTs adequately address the complex physical, emotional, and social needs of cancer survivors []. To address this gap, this systematic review aims to (1) examine how AI has been integrated into patient-facing DHTs designed to support cancer survivors, (2) narratively synthesize the potential effects of these technologies on HRQOL and provide preliminary quantitative estimates through an exploratory meta-analysis, and (3) explore broader changes in additional PROs (secondary aim).