Machine learning-based predictive models for the occurrence of behavioral and psychological symptoms of dementia: model development and validation

Abstract

The behavioral and psychological symptoms of dementia (BPSD) are challenging aspects of dementia care. This study used machine learning models to predict the occurrence of BPSD among community-dwelling older adults with dementia. We included 187 older adults with dementia for model training and 35 older adults with dementia for external validation. Demographic and health data and premorbid personality traits were examined at the baseline, and actigraphy was utilized to monitor sleep and activity levels. A symptom diary tracked caregiver-perceived symptom triggers and the daily occurrence of 12 BPSD classified into seven subsyndromes. Several prediction models were also employed, including logistic regression, random forest, gradient boosting machine, and support vector machine. The random forest models revealed the highest area under the receiver operating characteristic curve (AUC) values for hyperactivity, euphoria/elation, and appetite and eating disorders; the gradient boosting machine models for psychotic and affective symptoms; and the support vector machine model showed the highest AUC. The gradient boosting machine model achieved the best performance in terms of average AUC scores across the seven subsyndromes. Caregiver-perceived triggers demonstrated higher feature importance values across the seven subsyndromes than other features. Our findings demonstrate the possibility of predicting BPSD using a machine learning approach.

Using AI-Based Technologies to Help Nurses Detect Behavioral Disorders: Narrative Literature Review

“…A total of 8 publications addressed acute care settings [ 31 - 38 ], 6 looked at nursing homes [ 39 - 44 ], and 3 examined community care [ 45 - 47 ]. The majority described prospective observational studies published in journals covering geriatrics and psychogeriatrics [ 31 , 33 , 34 , 36 , 46 - 48 ] ( Multimedia Appendix 2 ) [ 31 - 55 ].…”

“…A total of 12 publications reported using machine learning–type AITs, including the facial expression recognition and predictive modeling subtypes [ 31 , 34 - 38 , 43 , 46 , 47 , 49 , 50 , 55 ], 11 publications explored digital health–type AITs, including the wearable technologies and robotic subtypes [ 32 , 39 - 42 , 44 , 45 , 51 - 54 ], and 2 publications examined natural language processing–type AITs [ 33 , 48 ] ( Figure 2 ).…”

“…Three publications referred to the use of natural language processing for screening and assessing clinical symptoms [ 35 , 48 , 49 ] and 1 reported on the use of machine learning [ 50 ]. One publication described the use of wearable technologies [ 40 ], and 1 combined this type of AIT with machine learning [ 46 ]. Finally, 10 publications reported on the use of robotics as a psychosocial approach to managing dementia and BPSD [ 32 , 39 , 41 , 42 , 44 , 45 , 51 - 54 ] ( Table 1 ).…”

A mobile app to predict and manage behavioral and psychological symptoms of dementia: Development, usability, and users’ acceptability

“…Several prediction models were employed using these data, including logistic regression, random forest, gradient boosting machine, and support vector machine. 29 Although these models showed different predictive performances across BPSD symptoms, the gradient boosting machine models showed the best performance in predicting BPSD, with a higher average area under the receiver operating characteristic curve (AUC) values across the symptoms (AUC > 0.8 except aberrant motor behaviors). This prediction function was installed in the mobile app in this developmemt.…”

“…In a previous study, the research team developed a machine-learning-based BPSD predictive model and validated its accuracy among PLWD. 29 The predictive model was developed using demographic and health information (e.g., age, sex, dementia diagnosis), functional status (ADL), premorbid personality, actigraphy data for sleep and physical activity, and BPSD diaries recorded for 2 weeks at baseline. Several prediction models were employed using these data, including logistic regression, random forest, gradient boosting machine, and support vector machine.…”