Update README.md

README.md

@@ -284,21 +284,6 @@ https://fdc.nal.usda.gov/
 
 ---
 
-## Related Research
-
-Both datasets were used in:
-
-> *Matrix Completion for Nutrient Imputation in Food Composition Databases*
-
-which evaluates SoftImpute, Masked NMF, KNN, and MICE imputation methods and demonstrates that matrix-factorization methods generalize across food composition databases of different scales and geographic origins.
-
-Key findings:
-- Both databases exhibit low-rank structure (rank 3 explains >66% variance), supporting the use of matrix completion.
-- SoftImpute is stable across 5%–50% missingness on both datasets.
-- The crossover point where SoftImpute outperforms KNN scales with database size (~25% for TKPI, ~35–40% for USDA SR Legacy).
-
----
-
 ## Repository File Structure
 
 ```