LSTM-CNN: An efficient diagnostic network for Parkinson's disease utilizing dynamic handwriting analysis.

BACKGROUND AND OBJECTIVES: Dynamic handwriting analysis, due to its noninvasive and readily accessible nature, has emerged as a vital adjunctive method for the early diagnosis of Parkinson's disease (PD). An essential step involves analysing subtle variations in signals to quantify PD dysgraphia. Although previous studies have explored extracting features from the overall signal, they may ignore the potential importance of local signal segments. In this study, we propose a lightweight network architecture to analyse dynamic handwriting signal segments of patients and present visual diagnostic results, providing an efficient diagnostic method. METHODS: To analyse subtle variations in handwriting, we investigate time-dependent patterns in local representation of handwriting signals. Specifically, we segment the handwriting signal into fixed-length sequential segments and design a compact one-dimensional (1D) hybrid network to extract discriminative temporal features for classifying each local segment. Finally, the category of the handwriting signal is fully diagnosed through a majority voting scheme. RESULTS: The proposed method achieves impressive diagnostic performance on the new DraWritePD dataset (with an accuracy of 96.2%, sensitivity of 94.5% and specificity of 97.3%) and the well-established PaHaW dataset (with an accuracy of 90.7%, sensitivity of 94.3% and specificity of 87.5%). Moreover, the network architecture stands out for its excellent lightweight design, occupying a mere 0.084M parameters, with only 0.59M floating-point operations. It also exhibits nearly real-time CPU inference performance, with the inference time for a single handwriting signal ranging from 0.106 to 0.220 s. CONCLUSIONS: We present a series of experiments with extensive analysis, which systematically demonstrate the effectiveness and efficiency of the proposed method in quantifying dysgraphia for a precise diagnosis of PD.

Transforming fatigue assessment: Smartphone-based system with digitized motor skill tests.

BACKGROUND: The condition of fatigue is a complex and multifaceted disorder that encompasses physical, mental, and psychological dimensions, all of which contribute to a decreased quality of life. Smartphone-based systems are gaining significant research interest due to their potential to provide noninvasive monitoring and diagnosis of diseases. OBJECTIVE: This paper studies the feasibility of using smartphones to collect motor skill related data for machine learning based fatigue detection. The authors' main goal is to provide valuable insights into the nature of fatigue and support the development of more effective interventions to manage it. METHODS: An application for smartphones running on Android OS is developed. Two aim-based reaction tests, an Archimedean spiral test, and a tremor test, were assembled. 41 subjects participated in the study. The resulting dataset consists of 131 trials of fatigue assessment alongside digital signals extracted from the motor skill tests. Six machine learning classifiers were trained on computed features extracted from the collected digital signals. RESULTS: The collected dataset SmartPhoneFatigue is presented for further research. The real-world utility of this database was shown by creating a methodology to construct a fatigue predictive model. Our approach incorporated 60 distinct features, such as kinematic, angular, aim-based, and tremor-related measures. The machine learning models exhibited a high degree of prediction rate for fatigue state, with an accuracy exceeding 70%, sensitivity surpassing 90%, and an f1-score greater than 80%. CONCLUSION: The results demonstrate that the proposed smartphone-based system is suitable for motion data acquisition in non-controlled environments and shows promise as a more objective and convenient method for measuring fatigue.

Why Behavioral Indicators May Fail to Reveal Mental States: Individual Differences in Arousal-Movement Pattern Relationships.

It is commonly assumed that behavior reflects the mental states of individuals. However, recent attempts to detect human states of mind via behavioral indicators have not always been successful; behavioral indicators may be unreliable and invalid. In this study we show that one of the common behavioral indicators, change in the overall amount of movement, correlated well with changes in the skin conductance level (SCL) at the group level, which reflects changes in arousal. At the individual level, however, changes in the SCL were related to movement patterns only in about half of the individuals. It is also noteworthy that the level of movement-SCL correlation was very highly predictable by certain social and cognitive characteristics of the individuals. Our results suggest that behavioral indicators may in many cases fail to predict mental states at the individual level.