UNSTRUCTURED DATA ANALYSIS FOR HEALTHIER PATIENT BEHAVIOR

Healthcare

THE CHALLENGE.

When patients take the medications they’re prescribed, it can have a major impact on their health, as well as on the success of the companies that produce pharmaceuticals and other treatments.

A biotechnology company wanted to use its case notes to understand why patients did or did not follow their medication regimens. But the descriptions of patient interactions were inaccurate, incomplete or in inconsistent formats, making valuable insights into their behavior difficult or impossible for the company to capture and act on.

UNSTRUCTURED DATA ANALYSIS FOR HEALTHIER PATIENT BEHAVIOR

Healthcare

THE CHALLENGE.

When patients take the medications they’re prescribed, it can have a major impact on their health, as well as on the success of the companies that produce pharmaceuticals and other treatments.

A biotechnology company wanted to use its case notes to understand why patients did or did not follow their medication regimens. But the descriptions of patient interactions were inaccurate, incomplete or in inconsistent formats, making valuable insights into their behavior difficult or impossible for the company to capture and act on.

Google TensorFlow™

APPROACH.

Working with the company to understand its products, patients and business needs, we identified the words and phrases of greatest interest within its case notes and built the ontologies and taxonomies required to train an artificial intelligence application to recognize this content. Our life sciences technology experts and the company applied machine learning and natural language processing (NLP) to years of unstructured, free text notes. To more effectively share the findings with client stakeholders and senior leadership, we created a 40-page narrative that presented our results in an understandable and actionable format.

RESULTS.

IMPRESSIVE INSIGHTS

Uncovered 23 meaningful insights and 11 key recommendations

360 COVERAGE

Partnered with stakeholders to create taxonomies and ontologies

Analysis

Developed KPIs to monitor and encourage actions that maximize patient wellness and drug sales

VALUE.

The insights from machine learning and NLP improved patient support, increased the odds of patients properly taking medications and identified roadblocks that caused them to stop treatment. The company created new key performance indicators (KPIs) for business processes, workflow improvements and coaching for improved patient engagement. Next steps include more complete documentation of the insights, training in documentation techniques and exploring how this approach could improve other functions, such as sales and marketing.