How We Used Text Mining and Intelligent Data Analytics in Patient Experience Transformation

Client: Large Hospital
Service: Intelligent Data Analytics

A general hospital in Canada, with about 300,000 patient visits a year noticed that they had mixed patient reviews on the Web and were looking to improve internal operations so that they can provide a better patient experience and start limiting negative comments.

They also needed to understand areas of excellence to replicate it in all areas of their practice. As their comments are mostly scattered around the Web and are highly unstructured in nature, it’s not easy for them to pick up on trend in complaints (or praises) by reading thousands of patient comments.

Our solution

Our first step was to understand the type of questions the hospital wants answered and to get an understanding of data that was available for analysis internally (if any) in addition to reviews on the Web.

Since privately managed data was limited, we gathered patient reviews from across the Web. Our crawler looked at sources such as RateMDs.com, HealthGrades.com and Vitals.com as these are online venues where patients leave detailed free-form feedback.

Once the data gathering pipeline was set-up, we aggregated all the reviews, preprocessed it and performed a series of text and sentiment analysis. We leveraged our very own state-of-the-art tools that uses Machine Learning and Natural Language Processing technologies, which was customized to answer questions that the hospital was interested in.

After our data analysis was complete (all questions answered), we summarized the results, separated signal from noise and delivered our findings. This document would be the key to making strategic decisions to provide a better patient experience.

Results

As a result of our partnership, the hospital gained several important insights that would enable them to drive operational changes. Some of the insights they obtained include:

  1. There were spikes in negative comments in recent months
  2. Specific issues were resulting in negative comments, specifically rude nurses and wait times
Sentiment Distribution Over the Years

Sentiment Distribution Across Themes

Key Themes within Negative Comments Related to “Timeliness”

Braced with this information, the hospital was able to take the next steps towards addressing their problems and investigate issues such as the cause for recent negative comments, which included staffing levels. The hospital aims to redo the analysis in subsequent quarters to look for improvements in specific aspects that we analyzed.

Need help on a similar problem? Get in touch to speak to our experts.
Scroll to Top