Latent dirichlet analysis based opinion mining of healthcare report

Abstract

Intelligent abstraction of knowledge from Social blogs has gained immense attention from Biomedical and online health communities. Healthcare re-views through public opinion platforms not only helped in improving the health-care but also reducing the costs. Advancement in digital healthcare systems enables patients access to Online Health Communities (OHC) through their views, opinions and remedial information. In the current study reviews provided by patients on OHC related to diabetes has been chosen to understand the community effort to address the health care issues of these patients. Reviews shared by patients at various levels of diabetic conditions have been selected and analyzed using LDA text mining techniques. In the current study we have also analyzed the gender specific differences among the diabetic community based on the opinions shared by them on public OHC platforms. Sample for the study has been gathered from a prominent online diabetic community in UK. Community members wrote on different topics from food habits, lifestyle, physical exercise and dis-ease symptoms to predict and possible remedial measures to be taken has been discussed in online health communities. Results of our study show that male users are information-centric than the female users, while female users are more emotionally attached as compared to males. Study also reveals several other findings related to community support and state of healthcare sector with reference to treatment, medication and facilities available to the community.