AI-Driven Mobile Health Technology: New Breakthroughs in Cancer Rehabilitation

Based on a 2024 preprint systematic scoping review (following the PRISMA-ScR framework), this article reviews the current application status of AI-enabled mobile health (mHealth) technology in cancer patient rehabilitation, explores key scenarios such as remote monitoring, personalized intervention, and clinical decision support, and analyzes its potential and challenges in improving rehabilitation continuity and patients' quality of life. Original source: OpenAlex/OSF, link: https://osf.io/2d5fh.

After cancer treatment, patients often face long-term side effects such as physical dysfunction and chronic fatigue. Traditional rehabilitation relies on regular follow-up visits, which has problems like uneven resource distribution, low compliance, and lack of rehabilitation continuity. The integration of AI and mHealth (smartphones, wearable devices, etc.) to build a patient-centered rehabilitation ecosystem brings possibilities to solve these dilemmas.

Technology Evolution: 1. Data collection and recording (electronic diaries, no intelligent analysis); 2. Connection and remote monitoring (real-time collection via wearables, manual analysis); 3. AI-driven personalized intervention (identify abnormalities, predict risks, automatically adjust plans).

Core Architecture: Multimodal data fusion layer (physiological, self-reported, medical records, etc.), intelligent analysis engine (deep learning and other algorithms), decision support system (generate clinical recommendations), user interaction interface (gamified design to enhance engagement).

1. Symptom Monitoring and Early Warning: For example, lymphedema after breast cancer surgery—AI provides early warning through activity patterns, skin conductance, etc.
2. Personalized Exercise Guidance: Dynamically adjust exercise intensity (e.g., reduce intensity during chemotherapy), and computer vision corrects posture in real time.
3. Cognitive Rehabilitation: Gamified training dynamically adjusts difficulty, and NLP identifies cognitive changes.
4. Psychosocial Support: AI chatbots provide 24/7 emotional support, and social algorithms match peer support.

1. Data and Algorithms: Sample bias (more young, highly educated patients), heterogeneous data makes it difficult to balance generalization and personalization.
2. Privacy and Security: Risk of sensitive data leakage—federated learning may be a solution.
3. Clinical Integration: Poor interoperability with electronic health records, alarm fatigue issues.
4. Regulatory Approval: Software as a Medical Device (SaMD) requires strict approval, and traditional clinical trials are hard to adapt to rapid iterations.

1. Multimodal Fusion and Digital Twins: Integrate genomic data to build patient digital twins, enabling 4P medicine (Predictive, Preventive, Personalized, Participatory).
2. Edge Computing: AI inference on the device side to reduce latency and enhance privacy.
3. Patient Empowerment: Transparent and explainable AI (XAI) helps patients participate in decision-making and enhances doctor-patient trust.

AI+mHealth redefines the boundaries of cancer rehabilitation—from passive to active, standardized to personalized, hospital-centered to patient-centered—with the potential to improve survivors' quality of life. However, it requires collaboration among technology, clinical practice, patients, and policy to address issues like privacy, integration, and equity, so that technology can benefit all patients.