An intelligent model in Human physical activity recognition for personal health assistant

Background and aims: Wearable technology has revolutionized the way we monitor and trackour daily activities. With the advent of advanced sensors, it has become possible to track andanalyze human behavior in real-time, providing valuable insights into our physical and mentalwell-being. With the development of advanced machine learning algorithms, it has becomepossible to identify a person’s specific activity and behavior accurately. This has opened up newavenues for healthcare professionals, clinicians, and caregivers to provide personalized care andtreatment. This study has found applications in various domains, such as healthcare, mental healthcare, elder care, and sports monitoring.Method: The WISDM dataset is a valuable resource for developing human activity recognitionmodels as it provides a wide range of accelerometer data collected from various devices. Theproposed model that aims to accurately detect common daily activities by separating the X and Yaxes has shown promising results in recognizing human activities. The Bayesian network, whichlearns the relationships between the features, provides a way to capture the complex dependenciesbetween different aspects of the data, while the two-dimensional hidden Markov model capturesthe temporal dependencies between the activities. By using these models in combination, theproposed approach provides a powerful and accurate tool for recognizing human activities, whichcan be applied in various fields, such as healthcare, sports, and entertainment.Results: The proposed approach achieves high accuracy in recognizing human activity, makingit useful for various applications. It can track not only the activity but also other characteristicssuch as health, lifestyle, and nutrition to help people achieve healthy lifestyles and independence.Automated assistance offered by this approach can benefit people who require ongoing care. Themethod’s ability to identify potential areas for improvement or intervention in individuals’ dailyactivities could significantly enhance their quality of life and contribute to a more efficient andcost-effective healthcare system. Overall, the approach has the potential to revolutionize the fieldof human activity recognition and bring significant benefits for individuals and society as a whole.Conclusion: The method described in the paper combines two probabilistic models, a Bayesiannetwork and a two-dimensional hidden Markov model, to accurately identify the nature of individualactivity. The approach has potential applications in various fields, including healthcare,where it could be used to track human activity and help people achieve healthy lifestyles and independence.The method could also be used to assist people who require ongoing care, such as theelderly or those with disabilities, by providing insights into their daily activities and identifyingpotential areas for improvement or intervention. Overall, the approach offers a promising avenuefor leveraging probabilistic models to improve human health and well-being.

Keywords:

Machine learning , sensor , human physical activity , recognition , smart devices