Machine learning prediction of suicide attempts in major depression: Feature selection and model development using national survey data.
Аннотация
BACKGROUND
The study aimed to explore the risk factors of suicidal attempts (SA) in patients with major depressive disorders (MDD).
METHODS
Cross-sectional analysis of 3247 MDD patients from the National Survey on Symptomatology of Depression (NSSD) was conducted, with data split 7:3 for training/validation. Boruta's Algorithm and Lasso screened predictors, while logistic regression and nomogram were used to identify independent risk factors of SA and visualize the overall impact of these factors on the SA risk of each patient.
RESULTS
392 (12.1 %) patients were found to have a history of SA. Boruta's Algorithm and Lasso analysis identified 20 variables as significant risk factors of SA, especially self-harm and a sense of decreased ability. Logistic regression analysis found that No. of hospitalization (OR = 1.14, 95 %CI:1.06, 1.23), SSRIs use(OR = 1.57,95 %CI:1.03,2.38), antipsychotics use (OR = 1.82,95 %CI:1.08,3.04), mood congruent psychosis (OR = 1.32,95 %CI:1.03,1.70), self-harm (OR = 70.90,95 %CI:45.70, 113.00), gastrointestinal system complaints (OR = 1.42,95 %CI:1.13, 1.78), weight gain (OR = 2.11,95 %CI:1.04, 4.29),the feelings of helplessness(OR = 1.47,95 %CI:1.11,1.95), unhappiness (OR = 1.52,95 %CI:1.14, 2.00) and derealization (OR = 1.33,95 %CI:1.02, 1.72) were independently and significantly associated with increased risk of SA in MDD patients. The prediction model showed robust performance, with an area under the curves (AUC) of the receiver-operator characteristics of 0.935 and 0.937 in the training set and validation set, respectively.
CONCLUSION
The findings may help develop assessment tools for suicide risk in MDD patients and provide clues for further mechanistic studies.
LIMITATIONS: Because of the study's cross-sectional design, causality could not be established between the predictors and SA, and the retrospective data collection approach may introduce recall bias.
Предметные рубрики MeSH
Подготовка и перевод публикации
Обработка статьи · этапы, ошибки и управление
Ошибки и ограничения
История · было → стало
Состояние обновляется каждые 5 секунд, пока блок открыт. Процент показывается только там, где известен объём работы. Пауза останавливает продолжение этапа; текущий запрос к источнику может завершиться.
Тип публикации указан по данным источника. Уровень доказательности автоматически не присваивается.