Early Schizophrenia Prediction Using Wearable Devices and Machine Learning

Wearable technology and machine learning algorithms are harnessed to advance early prediction and diagnosis of schizophrenia. Different machine learning algorithms were tested for their capability to categorize people at risk of or diagnosed with schizophrenia using a dataset made up of physiological and behavioral data gathered via wearable devices. Notably, K-nearest neighbors and random forest came up as the top-performing models, attaining high F1-scores, demonstrating their capacity to balance accuracy and recall. Support vector machine (SVM), AdaBoost, and gradient boosting all displayed competitive performance. The study emphasizes the value of feature selection and data preparation in improving model performance. By enabling early detection and customized treatment approaches, these findings show promise for revolutionizing schizophrenia diagnosis and intervention. Nevertheless, the selection of a machine learning algorithm should be in line with particular clinical aims, whether that means putting a focus on precision to cut down on false positives or recall to minimize missing instances. Although this study offers insightful information, additional validation on various datasets is necessary to see whether these models are generalizable.