CASE STUDY

Case study 1: Real-time patient risk stratification

We design innovative and impactful data-driven solutions for clients, which we deliver in the form of user-friendly applications and tools. We have extensive experience in implementing and assessing the impact of these solutions in partnership with our clients.

Services performed

  • Population health management consulting
  • Development of bespoke risk stratification model using a combination of clinical domain knowledge and unsupervised machine learning
  • Interactive visual analytics

Client

BrandMed is a connected health company in South Africa which owns an ecosystem of Bluetooth enabled wearable devices and point-of-care pathology machines. Readings from these devices and pathology machines are automatically stored in a cloud-based solution as an electronic record for each patient. 

Requirement

Having tested the devices in several practices and collected data from consenting patients, Mast Analytics was appointed to perform an analysis of the health-related information available. We assessed how the data collected by practices could be leveraged to support clinicians in understanding the clinical profiles of their patients at an individual and population level, to manage and improve patient care. 

Solution

We developed a bespoke risk stratification system that uses all the available information for each individual in the population. Unsupervised machine learning is used to assign each individual to a single, mutually exclusive risk cell. 

Our system includes a series of interactive dashboards that allow the clinician a deeper understanding of demographic and clinical risk factors associated with their patients, with a particular focus on understanding multi-morbidity to facilitate holistic treatment plans and improved patient care.