Using Maps As A Factor To Increase The Accuracy Of Collaborative Filtering In Providing Recommendations Regarding Cluster-Based Diseases Covid-19, Varicella And Dengue.

Abstract

In previous research, we classified disease types from a dataset of 66 thousand patient visits from 1/1/2019 to 12/31/2021 at Nadhifa Al Ghiffari, a health service institution in West Java, Indonesia. Using this data, we obtained data for diseases whose transmission is based on geographic clusters, namely Varicella (VAR), COVID-19 (COV) and Dengue Fever (DHF).

We tried to carry out several experiments using machine learning to classify types of disease and classification of referral/non refferal  types. In addition, we also conduct synthetic data experiments to increase the population of health data samples which are limited in number due to regulations related to medical confidentiality,

However, to strengthen accuracy, we also tried to process visualization of map data formed by the patient's address coordinates mapped with polygons in that area, compared with the location of old patients in the active period of transmission. This research shows what can be created with the data coordinates of patients  from the three diseases above.

All of this process is to provide early warning to health workers about the type of patient's disease, and whether to be referred or not, while from the government side it is necessary to observe the spread of the disease and the areas affected, in case special measures are needed such as isolating certain areas.