We at DatOS chose to "Hack for Health" and develop a three-pronged approach for the health and wellness space.
The first tool is a collaborative data access dashboard that allows data from multiple sources to be plugged in and analyzed together so Medlink can work directly with their partners.
The second tool was an unsupervised machine learning technique called clustering. We were able to identify new, unseen relationships within the data, and between neighborhoods in Charlotte. The Machine Learning technique grouped the neighborhoods into 4 clusters, or groups, based on their likeness in over 20+ attributes. This adds value for Medlink, because different neighborhoods can be marketed differently for possible health concerns that exist in their communities, or sending mailers to another cluster of neighborhoods that are more affluent and likely to support their community through donations.
The third tool is a supervised machine learning linear regression model. We used this model to determine what factors contribute to newborns being born underweight. We identified 5 factors that are associated with this concern. These 5 factors supported 25% of the variance in this life-affecting issue. More research can be done on these related factors, and this can be used to pool different non-profits together using existing resources effectively and identify different underlying causalities in their targeted communities.
Built With
- power-bi
- r
