As humans age beyond early adulthood, their physical and mental functions tend to slowly worsen over time. One of the most common sources of severe mental decline in older adults are neurodegenerative diseases, conditions characterized by the progressive loss of neurons in the brain or peripheral nervous system.
Past studies have found that the mental decline and memory loss associated with neurodegenerative diseases are often preceded by mental health-related symptoms, such as low mood, a lack of motivation, anxiety and irritability. So far, however, the early detection of neurodegenerative conditions based on the emergence of neuropsychiatric symptoms has proved challenging.
Researchers at Chongqing Medical University and the Chongqing Key Laboratory of Oral Diseases recently explored the potential of a new approach to predict the onset of cognitive decline, which combines biological samples with machine learning.
Their paper, published in Translational Psychiatry, highlights the potential of this approach for the large-scale screening of older adults and the identification of people who are more at risk of developing neuropsychiatric disorders or neurodegenerative diseases.
“Neuropsychiatric symptoms are early indicators of cognitive decline due to neurodegenerative diseases, and their timely detection is of the utmost importance,” wrote Ping Liu, Zeng Yang and their colleagues in their paper.
“We aimed to develop and validate methods for large-scale NPS screening among elderly individuals and explore underlying metabolic mechanisms.”
Analyzing saliva-related biomarkers with machine learning
To conduct their study, Liu, Yang and their colleagues recruited a total of 338 older adults who were accessing services at community health care centers in Chongqing, in China. These participants were asked to complete a questionnaire in which they shared their demographic information, while also providing samples of their saliva and of the bacteria inside their mouth.
The researchers also measured markers of the participants’ stress levels, such as the hormone cortisol and small proteins produced by immune cells called cytokines. The information they collected was then divided into two datasets.
The first contained the data collected from 138 of the participants and was used to train machine learning models. The second dataset, on the other hand, contained data taken from the remaining 200 participants and was used to validate the model’s ability to predict the risk that individual patients would experience neuropsychiatric symptoms.
The team developed and trained several different types of machine learning models, including a so-called Extreme gradient boosting (XGBoost), a support vector machine (SVM) and a logistic regression (LR) model. They then tested their ability to reliably identify patients who were at a higher risk of experiencing neuropsychiatric symptoms by analyzing biomarkers derived from the analysis of the saliva and oral microbiome samples.
They found that the XGBoost model performed better than other models. They then used one of the models they developed to create a platform that could be easily accessed by health care providers and used to screen groups of older adults.
“The genus-augmented XGBoost model achieved the highest performance, with an AUROC of 0.936 and an F1 score of 0.864, outperforming other models,” wrote the authors.
“The LR model was converted into a nomogram to facilitate neuropsychiatric-risk assessment in community settings. The external validation confirmed the strong predictive power (AUROC = 0.986, F1 score = 0.944). Enrichment and correlation analyses revealed cortisol and microbial interactions with pathways such as the pentose phosphate pathway and enterobacterial common antigen biosynthesis.”
Potential applications of the new AI-based tool
The new machine learning-based screening tool developed by this team of researchers could soon be improved further and tested in real-world clinical settings. In the future, it could help health care providers to detect the emergence of neuropsychiatric symptoms and possibly also cognitive decline early, planning therapeutic interventions and support strategies accordingly.
“The XGBoost-augmented model and nomogram offer promising tools for community-based NPS screening, while enrichment analysis provides insights into biological mechanisms,” wrote Liu, Yang and their colleagues.
The initial results collected by these researchers highlight the potential of machine learning models for the analysis of biological data and the early detection of neuropsychiatric symptoms.
Other neuroscientists and psychiatry researchers could soon draw inspiration from this study to develop additional AI-based platforms for the large-scale screening of older adults or other populations who might be at a higher risk of developing specific conditions.