Deep learning approaches show promise for predicting childhood malnutrition: A comparative study with traditional machine learning methods using survey data
Deepak Bastola, Yang Li · 2026
Read Deep learning approaches show promise for predicting childhood malnutrition: A comparative study with traditional machine learning methods using survey data by Deepak Bastola, Yang Li (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/plos-pone-0353969.txt.
Description
Bastola and Li, in a 2026 study, compare sixteen prediction algorithms for spotting malnourished children in Nepal, using national survey data rather than clinic visits. Finding underfed children door to door is slow, and remote areas are often missed. A model that flags risk from household survey answers could aim scarce help where it counts.
Stunting, wasting, and low weight were merged into one malnutrition label from a 2019 child survey. Deep learning, boosting, and classic methods were all tried and scored on ten measures, with weight on catching true cases, since missing a hungry child costs most. A tablet-shaped deep network, TabNet, won, likely thanks to its attention design. Mother's education, household wealth, and child's age were the strongest predictors, with place, vaccines, and meal frequency behind. The authors offer the framework as a template for other low-resource countries.
Catalog record
| CREATOR | Deepak Bastola, Yang Li |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 60993 |
| LICENSE | CC_BY |
| RIGHTS | CC BY |
| SOURCE | PLOS |