Original title: Serum BDNF and cognitive risk in maintenance hemodialysis: a machine learning study
Qiong Tang, Jiakun Tian, Xinyu Ying, Yanyun Zhang +3 · September 2026
People receiving long-term hemodialysis, a treatment that filters blood when kidneys fail, often develop memory and thinking problems. Doctors need better tools to spot this risk early. Researchers tracked 130 dialysis patients who started with healthy brain function to see who declined after one year. They checked each patient's starting blood levels of BDNFA protein in the blood that helps brain cells survive and grow., a brain protein linked to nerve health. They built computer programs to predict mental decline using routine medical records, then tested whether adding BDNF levels helped. Patients with lower BDNF were more likely to show cognitive decline. However, adding BDNF to the prediction programs only gave a small, uncertain boost in accuracy. BDNF might help fine-tune risk estimates alongside other tests, but it cannot predict decline by itself.
Imagine you are a coach picking a starting lineup for a big game. You have stats on every player, but you are not sure which stats actually matter, so you try adding one more piece of information—say, how many hours each player slept last night—and see if it helps you pick better. In this paper, the researchers did the same with people whose kidneys have failed: they already had a way to predict who might develop memory problems, and they tested whether adding one more blood test result made the predictions meaningfully better. It helped a little, but not enough to be sure it was not just luck.
Dialysis patients often develop memory and thinking problems, but doctors have few practical ways to predict who is at risk. This study tests whether a blood protein called BDNF adds useful information to existing risk scores. The findings suggest it may help refine estimates, but the improvement is small and needs confirmation in larger, diverse groups.
The study is single-center with 130 patients, the added predictive value of BDNF was small and not statistically significant, and the outcome was screening-defined rather than a full clinical diagnosis. Independent multicenter external validation is required before clinical use.
Ask anything about this paper. Answers come from its abstract, and say so when the abstract doesn't cover your question.
Serum BDNF and cognitive risk in maintenance hemodialysis: a machine learning study — Qiong Tang, Jiakun Tian, Xinyu Ying et al. — Renal failure — doi:10.1080/0886022x.2026.2732422
Abstract licence: CC BY
Explanation generated by Synthesis from the abstract — it is not the original paper.