Using Machine Learning and the HAMD-24 Scale to Predict Suicide Ideation in Depressed Patients
Bibliographic record
Abstract
Yun Chen,1 Zhong-Yi Jiang,2 Guan-Zhong Dong,1 Wei-Yuan Zhang,1 Ke Wang,3 Hai-Yan Yang1 1Department of Psychology, Nanjing Medical University Affiliated Changzhou Second People’s Hospital, Changzhou, Jiangsu, 213000, People’s Republic of China; 2School of Computer and Artificial Intelligence, Changzhou University, Changzhou, Jiangsu, 213000, People’s Republic of China; 3Nursing Teaching and Research Section, Nanjing Medical University Affiliated Changzhou Second People’s Hospital, Changzhou, Jiangsu, 213000, People’s Republic of ChinaCorrespondence: Hai-Yan Yang, Department of Psychology, Nanjing Medical University Affiliated Changzhou Second People’s Hospital, No. 68 Ge Lake Middle Road, Wujin District, Changzhou, Jiangsu, 213000, People’s Republic of China, Tel +86 0519-81099988, Email yanghaiyanp2@126.comObjective: The aim of this study was to identify factors associated with suicidal ideation and to develop a prediction model for early suicide ideation risk using machine learning algorithms based on the Hamilton Depression Scale (HAMD-24).Methods: A total of 374 patients with depression were included from the outpatient department of the Psychology Department at the Second People’s Hospital of Changzhou City. Depression severity was assessed using the HAMD-24, while the Beck Suicide Ideation (BSI) Questionnaire (Chinese Version) was employed to categorize patients into those with and without suicidal ideation. Suicide ideation risk in patients with depression was predicted using four machine learning models: support vector machine, naive Bayes classification, random forest, and extreme random trees classification (ERTC). This superiority is attributed to ERTC’s extreme randomization which mitigates overfitting in high-dimensional symptom data. The models were evaluated based on accuracy, precision, recall, F1 scores, Kappa coefficients, Matthew’s correlation coefficients, and area under the curve values. The optimal model was then selected, and the factors most strongly associated with suicidal ideation using the HAMD-24 were identified and analyzed.Results: The ERTC model outperformed SVM, NBC and RF (accuracy 77.75%, AUC 0.80), and despair, guilt, inferiority complex, work and interests loss, depression emotions were the strongest predictors of suicidal ideation. Demographically, patients with suicidal ideation were significantly younger and less likely to be using antidepressants. This is likely attributable to its ensemble structure and inherent randomization during node splitting, which enhances robustness against overfitting and improves generalization when handling the complex, potentially non-linear relationships between HAMD-24 items and suicidal ideation.Conclusion: We identified the optimal model and then analyzed the factors most strongly associated with HAMD-24 suicidal ideation. The ERTC model, demonstrating superior performance, enables early interventions, and reduces suicide rates. Moreover, this model provides a theoretical reference for the development of new scales focused on depression and suicide.Keywords: depression, machine learning, predictive model, suicidal ideation
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.004 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.001 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".