ESG Weight loss Study
2024
For this study, I worked with True You, a weight loss clinic based in North Carolina. Their team had done many Endoscopic Sleeve Gastroplasty (ESG) procedures on clients, and wanted to run an analysis to extract meaningful insights.
To do this, I ran a study exploring the relationship between personality traits and patient satisfaction with their surgery. This involved data transformation, of programmatically transcribing patient interviews to get text data, then using NLP to score patients on different personality metrics. Analysis revealed outcomes correlated primarily with initial weight rather than personality, guiding patient expectations and treatment approaches.
This data analysis posed a unique challenge. All the data of the patient text samples were interviews, not written text. To extract text from this, videos were transcribed using an API. Then I coded a tool to programmatically decipher which transcript was the patient’s and which was the doctor’s, and then automatically clean the data reformatted it. I then ran the Scaled Insights NLP AI model, extracting 113 personality traits for each patient’s transcript. Running a clustering analysis on the data, using both data of ESG patients, and True You's patients who pursued other treatment, results emerged. The personality based clusters had a significant difference in their rate of proceeding with the ESG.
My custom-made pipeline was able to transcribe, diarize, clean, transform, analyze, and visualize data in a streamlined process.
The client left with this knowledge of which traits lead to higher rates of satisfied patients. Additionally, we proved that the surgery was successful in reducing weight loss that scaled consistently to the patient's starting weight, and was not affected by any personality variables.
This project used a Python data pipeline for NLP-based personality scoring, clustering, demographics analysis, with p-value testing to prove statistical significance. Additionally, Microsoft suite was used, with Excel being used to format internal reports and visualize data, and with PowerPoint being used to summarize our findings to the client.
Comparitive Analysis
This survey also included a comparative analysis, comparing the True You patient database to some of the other data I had analyzed for Scaled Insights.
I had run 5 separate surveys at this point for various studies in the healthcare field. Some had been for smoking cessation, some for weight loss, and of varying scope and size.
The goal of the Comparative analysis was to see if True You had a distinct client base they were targeting.
As can be seen in the Combined PCA plots (2D & 3D) and the t-SNE plot, the TrueYou patient database (pink) has a very distinct personality profile from our other studies.
The ESG patients didn’t overlap much with other studies on weight loss, and were well clustered together.
