An official website of the United States government
Deep learning (DL) is of great interest in psychiatry due its potential yet largely untapped ability to utilize multidimensional datasets (such as fMRI data) to predict clinical outcomes. Typical DL methods, however, have strong assumptions, such as large datasets and underlying model opaqueness, that are suitable for natural image prediction problems but not medical imaging. Here we describe three relatively novel DL approaches that may help accelerate its incorporation into mainstream psychiatry research and ultimately bring it into the clinic as a prognostic tool. We first introduce two methods that can reduce the amount of training data required to develop accurate models. These may prove invaluable for fMRI-based DL given the time and monetary expense required to acquire neuroimaging data. These methods are (1) transfer learning − the ability of deep learners to incorporate knowledge learned from one data source (e.g., fMRI data from one site) and apply it toward learning from a second data source (e.g., data from another site), and (2) data augmentation (via Mixup) − a self-supervised learning technique in which “virtual” instances are created. We then discuss explainable artificial intelligence (XAI), i.e., tools that reveal what features (and in what combinations) deep learners use to make decisions. XAI can be used to solve the “black box” criticism common in DL and reveal mechanisms that ultimately produce clinical outcomes. We expect these techniques to greatly enhance the applicability of DL in psychiatric research and help reveal novel mechanisms and potential pathways for therapeutic intervention in mental illness.
more »
« less
This content will become publicly available on September 1, 2026
Harnessing Small‐Data Machine Learning for Transformative Mental Health Forecasting: Towards Precision Psychiatry With Personalised Digital Phenotyping
ABSTRACT Advances in digital phenotyping have opened the door to continuous, individualised monitoring of mental health, but realising the full potential of these data demands machine learning models that can operate effectively in ‘small‐data’ regimes—where per‐user data are sparse, irregular and noisy. This article explores the feasibility, challenges and opportunities of small‐data machine learning approaches for forecasting individual‐level mental health trajectories. We examine the limitations of traditional clinical tools and population‐level models and argue that fine‐grained time‐series forecasting, powered by models such as tabular prior‐data fitted networks (TabPFN), Gaussian processes, Kalman filters and meta‐learning strategies, offers a path towards personalised, proactive psychiatry. Emphasis is placed on key clinical requirements: real‐time adaptation, uncertainty quantification, feature‐level interpretability and respect for interindividual variability. We discuss implementation barriers including data quality, model transparency and ethical considerations and propose practical pathways for deployment—such as integrated biosensor platforms and just‐in‐time adaptive interventions (JITAIs). We highlight the emerging convergence of small‐data ML, mobile sensing and clinical insight as a transformative force in mental healthcare. With interdisciplinary collaboration and prospective validation, these technologies have the potential to shift psychiatry from reactive symptom management to anticipatory, personalised intervention.
more »
« less
- Award ID(s):
- 1922658
- PAR ID:
- 10649780
- Publisher / Repository:
- Wiley
- Date Published:
- Journal Name:
- Med Research
- Volume:
- 1
- Issue:
- 2
- ISSN:
- 2998-4963
- Page Range / eLocation ID:
- 226 to 238
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
In this mini-review, we discuss the fundamentals of using technology in mental health diagnosis and tracking. We highlight those principles using two clinical concepts: (1) cravings and relapse in the context of addictive disorders and (2) anhedonia in the context of depression. This manuscript is useful for both clinicians wanting to understand the scope of technology use in psychiatry and for computer scientists and engineers wishing to assess psychiatric frameworks useful for diagnosis and treatment. The increase in smartphone ownership and internet connectivity, as well as the accelerated development of wearable devices, have made the observation and analysis of human behavior patterns possible. This has, in turn, paved the way to understand mental health conditions better. These technologies have immense potential in facilitating the diagnosis and tracking of mental health conditions; they also allow the implementation of existing behavioral treatments in new contexts (e.g., remotely, online, and in rural/underserved areas), and the possibility to develop new treatments based on new understanding of behavior patterns. The path to understand how to best use technology in mental health includes the need to match interdisciplinary frameworks from engineering/computer sciences and psychiatry. Thus, we start our review by introducing bio-behavioral sensing, the types of information available, and what behavioral patterns they may reflect and be related to in psychiatric diagnostic frameworks. This information is linked to the use of functional imaging, highlighting how imaging modalities can be considered “ground truth” for mental health/psychiatric dimensions, given the heterogeneity of clinical presentations, and the difficulty of determining what symptom corresponds to what disease. We then discuss how mental health/psychiatric dimensions overlap, yet differ from, psychiatric diagnoses. Using two clinical examples, we highlight the potential agreement areas in assessment/management of anhedonia and cravings. These two dimensions were chosen because of their link to two very prevalent diseases worldwide: depression and addiction. Anhedonia is a core symptom of depression, which is one of the leading causes of disability worldwide. Cravings, the urge to use a substance or perform an action (e.g., shopping, internet), is the leading step before relapse. Lastly, through the manuscript, we discuss potential mental health dimensions.more » « less
-
To develop efficient just-in-time personalised treatments, dynamical models are needed that provide a description of how an individual responds to treatment. However, available system identification approaches cannot effectively be applied to most behavioural datasets since, usually, the data collected is subjected to a large amount of noise and time sampling is not uniform. To be able to circumvent these issues, in this paper a new method is proposed for parsimonious system identification of continuous-time systems that does not require specially structured data. The developed algorithm provides an effective way to leverage these ‘non-standard’ datasets to identify continuous time dynamical models that are compati- ble with a-priori information available on the process. The algorithm developed is tested on data obtained from a behavioural study on adolescents and violence. The objective is to model the temporal dynamics of the association between violence exposure and mental health symptoms (depression and anxiety) in day- to-day life among a sample of adolescents at heightened risk for both substance use exposure and problem behaviour. The information extracted from individual models of behaviour such as the maximum burden and the time of fading away of depression/anxiety does differ substantially from person to person. This information has the potential to be useful to design personalised interventions that would have a better chance of succeeding.more » « less
-
Transformer model architectures have revolutionized the natural language processing (NLP) domain and continue to produce state-of-the-art results in text-based applications. Prior to the emergence of transformers, traditional NLP models such as recurrent and convolutional neural networks demonstrated promising utility for patient-level predictions and health forecasting from longitudinal datasets. However, to our knowledge only few studies have explored transformers for predicting clinical outcomes from electronic health record (EHR) data, and in our estimation, none have adequately derived a health-specific tokenization scheme to fully capture the heterogeneity of EHR systems. In this study, we propose a dynamic method for tokenizing both discrete and continuous patient data, and present a transformer-based classifier utilizing a joint embedding space for integrating disparate temporal patient measurements. We demonstrate the feasibility of our clinical AI framework through multi-task ICU patient acuity estimation, where we simultaneously predict six mortality and readmission outcomes. Our longitudinal EHR tokenization and transformer modeling approaches resulted in more accurate predictions compared with baseline machine learning models, which suggest opportunities for future multimodal data integrations and algorithmic support tools using clinical transformer networks.more » « less
-
Language models have the potential to assess mental health using social media data. By analyzing online posts and conversations, these models can detect patterns indicating mental health conditions like depression, anxiety, or suicidal thoughts. They examine keywords, language markers, and sentiment to gain insights into an individual’s mental well-being. This information is crucial for early detection, intervention, and support, improving mental health care and prevention strategies. However, using language models for mental health assessments from social media has two limitations: (1) They do not compare posts against clinicians’ diagnostic processes, and (2) It’s challenging to explain language model outputs using concepts that the clinician can understand, i.e., clinician-friendly explanations. In this study, we introduce Process Knowledge-infused Learning (PK-iL), a new learning paradigm that layers clinical process knowledge structures on language model outputs, enabling clinician-friendly explanations of the underlying language model predictions. We rigorously test our methods on existing benchmark datasets, augmented with such clinical process knowledge, and release a new dataset for assessing suicidality. PKiL performs competitively, achieving a 70% agreement with users, while other XAI methods only achieve 47% agreement (average inter-rater agreement of 0.72). Our evaluations demonstrate that PK-iL effectively explains model predictions to clinicians.more » « less
