This study explores the use of machine learning models to assess the severity of Parkinson’s disease symptoms based on data from wearable and smartphone sensors. It presents models to predict the severities of individual symptoms—tremor, bradykinesia, stiffness, and dyskinesia—as well as the overall state of patients, using both clinician and patient self-assessments as labels. The dataset, although limited and imbalanced, enabled the identification of key trends. The best performance was achieved when combining data from both the MYO armband and smartphone, and when using patient self-assessments as targets. Tremor was the most predictable symptom, while others proved more challenging—especially at higher severity levels, which were poorly represented in the dataset. These results highlight the value of multimodal data and the importance of patient input in symptom monitoring. However, they also point to the need for more balanced and extensive datasets to improve prediction accuracy across all severity levels and symptoms.

This paper presents a solution for creating individualized medicine intake schedules for Parkinson’s disease patients. Dosing medicine in Parkinson’s disease is a difficult and a time-consuming task and wrongly assigned therapy affects patient’s quality of life making the disease more uncomfortable. The method presented in this paper may decrease errors in therapy and time required to establish a suitable medicine intake schedule by using objective measures to predict patient’s response to medication. Firstly, it demonstrates the use of machine learning models to predict the patient’s medicine response based on their state evaluation acquired during examination with biomedical sensors. Two architectures, a multilayer perceptron and a deep neural network with LSTM cells are proposed to evaluate the patient’s future state based on their past condition and medication history, with the best patient-specific models achieving R² value exceeding 0.96. These models serve as a foundation for conventional optimization, specifically genetic algorithm and differential evolution. These methods are applied to find optimal medicine intake schedules for patient’s daily routine, resulting in a 7% reduction in the objective function value compared to existing approaches. To achieve this goal and be able to adapt the schedule during the day, reinforcement learning is also utilized. An agent is trained to suggest medicine doses that maintain the patient in an optimal state. The conducted experiments demonstrate that machine learning models can effectively model a patient’s response to medication and both optimization approaches prove capable of finding optimal medicine schedules for patients. With further training on larger datasets from real patients the method has the potential to significantly improve the treatment of Parkinson’s disease.

The paper presents a novel sensor-based disease symptoms evaluation method which can be applied in the domain of neurological treatment monitoring and efficiency analysis. The main purpose of the method is to provide a quantitative approach for symptoms recognition and their intensity, which can be used for efficient medicine intake planning for Parkinson's Disease patients. This work presents an innovative method, which enables to objectify the process of clinical trials. The developed solution implements sensor data fusion method, which analyses time correlated wearable sensor biomedical data and symptoms survey. We have merged two separate methods of recognizing and assessing the intensity of Parkinson's Disease (PD) symptoms using time-constrained survey as well as sensor and interaction-based algorithms, which enable to objectively assess the intensity of disease symptoms. Based on process-based analysis and clinical trials observations, a set of requirements for validating symptoms of neurological diseases have been formulated. Proposed solution concentrates on PD indicators connected with arms movement and mental reaction delays, which can be registered using wearable sensors. Since 2017 the tool has been tested by a group of four selected neurologists and 10 users, 3 of which are PD patients. To meet the project's supplementary (efficiency, security) requirements, a test clinical trial has been performed involving 3 patients executing trials which lasted two weeks and was supported by the continuous application usage. After successful deployment the method and software tools has been presented for commercial use and further development in order to adjust its usage for other neurological disorders.