Evidence Explorer · MS-N220 · Nutrition support and digital delivery · Appraised November 3, 2026
A small randomized trial found better short-term energy and protein target achievement with AI-facilitated, clinician-reviewed, individualized discharge guidance. It tested a multicomponent service—not AI alone—and did not improve weight, laboratory nutrition markers or postoperative complications at 30 days.
Evidence at a glance
Benefit signal: Positive intermediate dietary-intake outcome
Certainty: Low-to-moderate for 30-day target achievement; uncertain clinical benefit
Risk context: Clinician review was built in; unreviewed AI advice was not tested
What was studied
A single-center trial randomized 90 adults after laparoscopic gastric or colorectal cancer surgery, 45 per group. The intervention used DeepSeek v3.2, a structured prompt, individual clinical data, quantified targets, written guidance and dietitian safety review; control patients received standard generic discharge advice.
At 30 days, 66.7% versus 40.0% achieved at least 90% of their energy target (OR 3.0, 95% CI 1.3–7.1), and 62.2% versus 35.6% achieved the protein target. Complications were 20.0% versus 22.2%; weight and laboratory nutrition markers did not differ.
What this evidence cannot establish
The trial cannot separate AI from individualization, quantified goals, longer written material, clinician review or extra attention. Follow-up was 30 days and all participants were from one hospital.
The article’s symptom reporting contains internally inconsistent values for some domains. This appraisal therefore centers on the prespecified energy-adequacy endpoint and does not claim a reliable symptom benefit.
Practical safety context
Every AI-generated plan received dietitian review. The study does not support using a chatbot as an autonomous dietitian, especially with renal, hepatic, swallowing, ostomy or dumping-syndrome concerns.
The intervention did not establish fewer complications, better recovery, recurrence control or survival.
Questions for the care team
- Who reviews the plan for medical and food-safety errors?
- Are my calorie, protein, fluid and micronutrient targets individualized?
- How will weight, intake, tolerance and laboratory abnormalities be monitored after discharge?
Sources
- Open-access randomized trial published October 29, 2026 (DOI 10.1007/s00520-026-11401-7)
Sources and publication status checked November 3, 2026. Full text was appraised, including the registered design, funding statement and reported outcome tables.
For education, not personal medical advice. Nutrition and supplement decisions during cancer care should be individualized with the oncology and dietetics teams.
