Vulnerable population in healthcare refer to those who are at greater risk of suffering from health hazards due to various socio-economic factors, geographical barriers and medical conditions. Mapping of this vulnerable population is a vital part of healthcare planning for any region. Results of vulnerability mapping later can help with meaningful interventions for healthcare demands.
This study focuses on combining geo analytics, unsupervised machine learning algorithm and entropy method for performing vulnerability mapping based on various above- mentioned factors.
In this study, k-means++ clustering algorithm is applied to household data of Ratnanagar municipality. Out of the available data, specific vulnerability indicators related to income, age, illness, disability and geolocation are created for the purpose of creating multiple clusters of households. One of the features used is, distance to the nearest health service provider, which is computed by using a routing engine called Open Source Routing Machine (OSRM), based on geolocations of households and health service providers of Ratnanagar. OpenStreetMap route data is used for this purpose.
After the clusters are formed, entropy method is used to evaluate vulnerability measure of each cluster. Later, based on population of different clusters in each ward and their respective vulnerability measures, each ward’s vulnerability measure is quantified. Finally, a ward level vulnerability map is created for the municipality.
The results of this research can help decision makers perform evidence-based healthcare planning. The decision makers can identify the most vulnerable areas of the municipality and take rationale based mitigative measures.
Keywords: healthcare, vulnerability mapping, k-means++ clustering, elbow method, entropy method, OpenStreetMap, open source routing machine