Table 7.

Features selected commonly and individually by the best three machine learning models.

Feature categories	Selected by 3 models	Selected by 2 models	Selected by 1 model
Demographic		Race	EthnicitySexAge
Anthropometric	Lean body massConicity index	Weight in kgWaist circumference in cmBia resistanceFat free mass	Height in cmBMI
Disease-Related	Length of dialysis treatmentDialysis accessDiabetes medication, insulinAnti-inflammatory medicationEtiology of CKD (hypertension)	Oral diabetes medicationAnti-lipemic medication	History of diabetesAnti-hypertensive medicationDialysis treatment timeEtiology of CKD (diabetes)
Clinical/Dynamic	Interdialytic weight gainHeart rate (bpm)	Weekly mean kt/vBody temperatureWeekly mean npcr	Weekly mean urrDystolic blood pressureSystolic blood pressure
Patient-Reported	Enjoy mealtimesWeekly appetite evaluationFeel like eatingDaily appetite rating	Intake on non-dialysis dayEnjoyment of eating on dialysis day	Difficulty following dialysis dietIntake on dialysis dayOverall appetite ratingAppetite on non-dialysis dayLack of appetite
Provider-Assessed	SGA physical examinationSGA gastrointestinal examinationSGA functional examination	SGA weight change assessmentSGA overall malnutrition rating	SGA disease assessment

bpm: beats per minute; BMI: body mass index; CKD: chronic kidney disease; kt/v: fractional urea clearance; npcr: normalized protein catabolic rate; SGA: Subjective Global Assessment; urr: urea reduction rate.