Truace
Subscribe

Model predicted psychiatric readmission within 90 days for 322 adults with depression

Study title: Predicting early psychiatric readmission among people with major depressive disorder: A machine learning analysis from the prospective multicentre DEEP READ study

The DEEP READ prospective study across 13 Italian provinces followed 322 adults with major depressive disorder discharged from inpatient care and tested whether routinely collected clinical information could predict unplanned psychiatric readmission within 90 days. A Random Forest classifier trained on 22 predictors achieved a mean test AUC of 0.74 in internal cross-validation, with 50 patients (15.5%) readmitted.

Read the study abstract

Background Major depressive disorder (MDD) is associated with high relapse rates, with around 20% of inpatients experiencing readmission within 90 days after discharge. Routinely collected clinical information may help identify individuals at higher risk of readmission. Machine learning (ML) models could complement more traditional statistical approaches by exploring complex relationships among multiple vulnerability factors. Methods The DEEP READ study is a prospective, multicentre cohort study conducted across…

By the Truvace desk1 min read1 sourceEvidence: Moderate · 60/100 — Observational study · n = 322 · 1 independent sourcePeer-reviewedTRV-2026-1096· cite & verify
Predicting early psychiatric readmission among people with major depressive disorder: A machine learning analysis from the prospective multicentre DEEP READ study
Thirty-second annual report of the Government Hospital for the Insane to the Secretary of the Interior. 1887 by Government Hospital for the Insane (U.S.) Toner, Joseph M. (Joseph Meredith), 1825-1896 Godding, W. W. (William Whitney), 1831-1899 Royal College of Physicians of London. Public domain
Trace impact readingContested
P 57The P score combines the specificity and measured human impact of the grounded problem claim with the strength of this Trace’s cited sources.

Both sides are scored from claims and sources, not community votes.

G 56The G score combines the specificity and measured human impact of the grounded gain claim with the strength of this Trace’s cited sources.

In brief

Moderate discrimination from standard variables suggests potential for risk stratification, but wide fold variation and lack of external validation leave clinical usefulness unproven. The authors note that calibration, generalizability beyond the Italian cohort, and impact on care decisions remain to be established before any implementation.

Main points

  1. Prospective multicentre cohort across 13 Italian provinces enrolled 322 adults aged 18-65 with DSM-5-TR MDD between January 2024 and December 2025.
  2. Primary outcome was unplanned psychiatric readmission within 90 days, observed in 15.5% (n = 50) of the sample.
  3. Leading predictors included suicide attempts, prior hospitalisations, anticonvulsant prescription, cluster B personality disorder, female sex, and anxious distress.

The gain

A Random Forest trained on 22 routine clinical variables predicted unplanned 90-day psychiatric readmission in adults with MDD with moderate discrimination.

The problem

The model showed variable cross-validation performance down to 0.59 AUC and is not validated for clinical implementation without further calibration and utility assessment.

The rundown

DEEP READ enrolled 322 individuals (mean age 43.0 years; 38.5% men) with DSM-5-TR MDD and recorded sociodemographic features, Hamilton Depression Rating Scale scores, DSM-5 specifiers, comorbidities, and pharmacological treatment at discharge.

The Random Forest used 22 predictors; SHAP analysis highlighted suicide attempts and prior hospitalisations as top contributors, while lithium prescription showed a negative SHapley Additive exPlanations (SHAP) pattern described as exploratory.

What this doesn’t fix

Model was only internally evaluated with stratified 5-fold cross-validation and lacks external validation, calibration assessment, and demonstrated clinical utility.

Sources

  1. Peer-reviewedJournal of Affective Disorders2026-09-13

The debate