Daily Human behaviors Based Personalized health Recommendation

Chaw, Moe Ni Ni (2025) Daily Human behaviors Based Personalized health Recommendation. Masters thesis, Dublin, National College of Ireland.

Preview	PDF (Master of Science) Download (935kB) | Preview
Preview	PDF (Configuration Manual) Download (378kB) | Preview

Abstract

Unhealth eating habits, poor sleep and lack of physical exercise have been causing the increase in diabetes, obesity and cardiovascular disease. The traditional health systems are weak in addressing such lifestyle related problems because these cannot provide to fit the individual needs. Moreover, the latest health tracking devices has limitations, and they often fail to take impact of daily behaviors such as drinking habit, meal timing and unstructured exercise on overall health and wellbeing. The most existing health system based on AI are rule based systems and based on the use of fixed recommendation which cannot adapted to daily behavior of individual. To address these limitations, a customize label dataset was created from fit-bit wearable device. The data were collected formed based on real life measurements obtained through wearable which contain various features. Each simple data was labeled based on medical and behavioral health standards to detect unhealth behaviors. These identified features are used as input data to multiple Long Short Term Memory architectures such as simple LSTM, Deep LSTM and Bidirectional LSTM and trained with unhealthy habits using sequential patterns of daily activities. The proposed Deep LSTM model achieved accuracy of 89 percent which is the highest classification result. And then, Reinforcement Learning algorithm is developed to generated personalized and adaptive health advice using the result of deep LSTM classification and trained to enhance recommendation performance using simulated user feedback which achieved 16 percent of mean improvement rate. This finding has shown that the integration of LSTM based identification of habits and RL based adaptive recommendation can be healthcare interventions. It has potential to reduce the risk of chronic diseases and improve health outcomes by overcoming the known drawbacks of static recommendation systems.

Item Type:	Thesis (Masters)
Supervisors:	Name Email Anand, Devanshu UNSPECIFIED
Subjects:	R Medicine > RA Public aspects of medicine > RA0421 Public health. Hygiene. Preventive Medicine Q Science > QH Natural history > QH301 Biology > Methods of research. Technique. Experimental biology > Data processing. Bioinformatics > Artificial intelligence Q Science > Q Science (General) > Self-organizing systems. Conscious automata > Artificial intelligence R Medicine > RA Public aspects of medicine > RA790 Mental Health R Medicine > RA Public aspects of medicine > RA0421 Public health. Hygiene. Preventive Medicine > Personal Health and Hygiene
Divisions:	School of Computing > Master of Science in Artificial Intelligence
Depositing User:	Ciara O'Brien
Date Deposited:	12 Aug 2026 08:37
Last Modified:	12 Aug 2026 08:37
URI:	https://norma.ncirl.ie/id/eprint/9501

Actions (login required)

View Item