Multimodal Intelligent Monitoring of Parkinson Disease: Scoping Review of Progress and Translational Challenges

Background: Parkinson disease (PD) is a progressive neurodegenerative disorder with a rapidly growing global prevalence. Current clinical assessments, such as the Unified Parkinson Disease Rating Scale, are limited by subjectivity and episodic application, creating a need for continuous, objective monitoring solutions. While previous reviews have often focused on single technologies, there is a growing trend toward integrating multiple data sources to provide a more holistic view of PD. Objective: This scoping review synthesizes progress in multimodal intelligent monitoring systems for PD, focusing on the quantification of motor and nonmotor symptoms, algorithm development, and the clinical translation of remote monitoring platforms. Furthermore, we propose a novel heuristic framework (Care-Platform Transformation in PD [CPT-PD]) that provides a forward-looking conceptual design for integrating these technologies into clinical workflows, demonstrating promising potential for future development. Methods: A targeted literature search was conducted on August 15, 2025, in PubMed, Web of Science, and China National Knowledge Infrastructure for research published between January 1, 2019, and December 31, 2024. The final search was rerun on January 22, 2026, solely to ensure completeness of coverage for this time window; no articles published after December 31, 2024, were included. Results: Wearable sensors (n=9) demonstrated high concordance with clinical scores in validation studies (eg, 99% for tremor detection), while computer vision (n=6) achieved moderate agreement with clinician ratings in controlled assessments (intraclass correlation coefficient 0.74 for bradykinesia). For nonmotor symptoms, intelligent systems (n=7) demonstrated sleep disturbance detection with up to 92.9% accuracy and autonomic dysfunction monitoring (n=7) via heart rate variability (area under the curve 0.90) and voice analysis (94.55% accuracy). Algorithm studies (n=16) explored single-modality feature extraction and cross-modal fusion, with emerging applications in federated learning. Remote platforms (n=22) improved medication adherence (172/201, 85.6%) and reduced outpatient visits (by 29% in one study). A heuristic CPT-PD framework was proposed to integrate key components of diagnosis, treatment, and management. Collectively, these advancements demonstrate the technical viability and clinical benefits of shifting from episodic, subjective assessments toward a data-driven, continuous, and multimodal approach to PD management. Conclusions: While current evidence largely reflects multisensor systems rather than deeply integrated multimodal platforms, the field holds promise for advancing toward genuine data fusion that could further improve clinical decision-making. Persistent challenges include fragmented symptom focus, algorithmic heterogeneity, and barriers to adoption among older adults. Future efforts should build on integrated frameworks such as CPT-PD to develop patient-centered ecosystems, ultimately enabling precision medicine in PD management.