Patient Weight Gain Narratives for Obesity Phenotyping, 2463 Patients

2463 patients with overweight or obesity provided open-ended weight gain cause narratives prior to a multidisciplinary weight loss treatment. The narratives were labeled using 12 thematic categories via a GPT4.1 large language model, and associations with demographic factors and treatment outcomes were analyzed. The dataset was authored by Miksa M. Henkrich and last updated on April 14, 2026.

Use Cases

• Automated thematic labeling of unstructured patient narratives based on 12 predefined categories.
• Predicting weight loss outcomes based on reported weight gain causes and demographic features.
• Identifying patient phenotypes via unsupervised clustering of narrative themes, age, sex, and BMI.
• Analyzing co-occurrence patterns and risk ratios between different reported weight gain causes.

Strengths

• 2463 patient narratives provide a substantial corpus for analysis.
• Automated labeling achieved precision of 0.906 and recall of 0.897 against a reference sample.
• Mean weight loss of 9.2% over 108.6 days provides a concrete outcome metric.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Data may reflect bias inherent to the specific clinical cohort studied.

Provenance

Source

figshare

Collection Method

Narratives collected from patients prior to a multidisciplinary medical-nutritional weight loss treatment.

Freshness

Last updated 2026-04-14 16:01:31; freshness should be verified.