A revolution is underway in the world of healthcare. Data lies at the center of it. Establishing meaningful use of EHRs, demonstrating quality care to upcoming ACOs, and cutting costs in chronic disease management in order to meet new healthcare mandates, all require advanced analytics. At Austin Labs, we are committed to providing healthcare organizations and business associates the best data science to fulfill the promises of this revolution.


Clinical Inferencing

Use EHR clinical data, claims data, prescription and labs history to predict outcomes such as expensive hospital revisits, risks of relapse. Clinical notes contain a wealth of information about patients’ medical history, present conditions, undergone procedures, and prescribed medications. Providers can leverage millions of available labeled and unlabeled clinical notes from newly mandated EHR systems to gain new levels of productivity and gain unprecedented level of insights.

Austin Labs participated in this $3M Kaggle data science competition to help Heritage Health Foundation identify the best predictors for re-hospitalization. Austin Labs was placed in the top 3% of the participants, our error score being within 1% of the leader. We used a custom ensemble of a number of different machine-learning techniques to achieve that rank in 37 attempts in comparison to 555 attempts made by the leading team.



Cost Outlier Analytics

Almost 80% of all healthcare costs in US stem from just 20% of patients. These patients also often suffer from complex and/or chronic diseases that require expensive and frequent medical treatment. An innovative idea being implemented in various parts of the country is to create centers of excellence that specialize in dealing with such diseases. These centers identify treatment paths that are most effective and execute them in the most efficient fashion and thus reducing overall costs. What’s still lacking however is a meaningful and rapid identification of patients whose care costs are approaching questionable levels? What’s even more challenging is to identify among these patients those whose costs have risen without any medical justification.

Use Cases and Benefits

  • Dynamically identifying when a patient’s treatment path has crossed into an outlying zone. Identify among those the subset whose costs have risen without an economic or medical justification.
  • Definition of outlying zones for targeted single or multiple morbidity patient populations
  • Definition of cost models that allow comparison across providers
  • Hyper-local definition of outliers allows a six-sigma style continuous improvement in costs at the right level of locality – hospital, zip-code, or state.

Population Health Analytics

Factors that put a person most at risk for initially developing a medical condition and then later potentially developing complications and even succumbing to it are best determined in the context of a particular morbidity, geography, and demographics. While genetics and specific prior medical conditions no doubt play a big role, we focus on identifying those risk factors that are more related to the person’s environment, demographics, and care quality the person received in the early phase of the medical condition. In doing so, we identify risk factors that are particular relevant for a specific patient receiving care in a specific clinic or hospital, making the insights immediately more actionable.


  • Dynamically determine in a statistically-valid way the top demographics, medical care/quality and behavioral factors behind incidences and progression of major diseases and medical conditions.
  • Hyper-local analytics to enable patient-level intervention
  • Hard, quantified indices to enable monitoring changes and improvement in risk factors