Testing domain knowledge and risk of bias of a large-scale general artificial intelligence model in mental health.

Background: With a rapidly expanding gap between the need for and availability of mental health care, artificial intelligence (AI) presents a promising, scalable solution to mental health assessment and treatment. Given the novelty and inscrutable nature of such systems, exploratory measures aimed at understanding domain knowledge and potential biases of such systems are necessary for ongoing translational development and future deployment in high-stakes healthcare settings. Methods: We investigated the domain knowledge and demographic bias of a generative, AI model using contrived clinical vignettes with systematically varied demographic features. We used balanced accuracy (BAC) to quantify the model's performance. We used generalized linear mixed-effects models to quantify the relationship between demographic factors and model interpretation. Findings: We found variable model performance across diagnoses; attention deficit hyperactivity disorder, posttraumatic stress disorder, alcohol use disorder, narcissistic personality disorder, binge eating disorder, and generalized anxiety disorder showed high BAC (0.70 ≤ BAC ≤ 0.82); bipolar disorder, bulimia nervosa, barbiturate use disorder, conduct disorder, somatic symptom disorder, benzodiazepine use disorder, LSD use disorder, histrionic personality disorder, and functional neurological symptom disorder showed low BAC (BAC ≤ 0.59). Interpretation: Our findings demonstrate initial promise in the domain knowledge of a large AI model, with performance variability perhaps due to the more salient hallmark symptoms, narrower differential diagnosis, and higher prevalence of some disorders. We found limited evidence of model demographic bias, although we do observe some gender and racial differences in model outcomes mirroring real-world differential prevalence estimates.

Process Evaluation of Two Interventions to Improve Health Environments in Family Child Care Homes.

Early care and education (ECE) environments influence children's lifelong health behaviors, growth, and development. Although the number of interventions to improve health in ECE environments is increasing, few have been designed for and tested in family child care homes (FCCHs). This study reports the process evaluation of two interventions to improve FCCH health environments, both part of Happy Healthy Homes, a matched-attention randomized-controlled intervention trial conducted in Oklahoma FCCHs. Participating child care providers received one of two multicomponent interventions: (a) an intervention focused on enhancing the quality of the nutrition environment, self-efficacy, and practices or (b) an intervention focused on improving providers' environmental health literacy, self-efficacy, and practices. Guided by "Implementation" constructs of the RE-AIM framework (i.e., reach, effectiveness, adoption, implementation, maintenance), intervention report forms and participant tracking were used to assess intervention dose delivered and interventionist-perceived challenges and successes to implementation. Interviews were conducted to obtain participant feedback after the intervention. Dose delivered was high for both interventions overall and across individual sessions, and provider feedback was positive. Implementation challenges and strengths identified for both interventions may be useful for further enhancing intervention appropriateness and success, particularly for interventions with FCCHs. Process evaluation findings indicate that the two Happy Healthy Home interventions can be conducted with high delivery and are well attended and considered to be valuable to FCCH providers.

Digital biomarkers of anxiety disorder symptom changes: Personalized deep learning models using smartphone sensors accurately predict anxiety symptoms from ecological momentary assessments.

Smartphones are capable of passively capturing persons' social interactions, movement patterns, physiological activation, and physical environment. Nevertheless, little research has examined whether momentary anxiety symptoms can be accurately assessed using these methodologies. In this research, we utilize smartphone sensors and personalized deep learning models to predict future anxiety symptoms among a sample reporting clinical anxiety disorder symptoms. Participants (N = 32) with generalized anxiety disorder and/or social anxiety disorder (based on self-report) installed a smartphone application and completed ecological momentary assessment symptoms assessing their anxiety and avoidance symptoms hourly for the course of one week (T = 2007 assessments). During the same period, the smartphone app collected information about physiological activation (heart rate and heart rate variability), exposure to light, social contact, and GPS location. GPS locations were coded to reveal the type of location and the weather information. Personalized deep learning models using the smartphone sensor data were capable of predicting the majority of total variation in anxiety symptoms (R2 = 0.748) and predicting a large proportion of within-person variation at the hour-by-hour level (mean R2 = 0.385). These results suggest that personalized deep learning models using smartphone sensor data are capable of accurately predicting future anxiety disorder symptom changes.

Polarization-resolved single-molecule tracking reveals strange dynamics of fluorescent tracers through a deep rubbery polymer network.

Tracking the movement of fluorescent single-molecule (SM) tracers has provided several new insights into the local structure and dynamics in complex environments such as soft materials and biological systems. However, SM tracking (SMT) remains unreliable at molecular length scales, as the localization error (LE) of SM trajectories (∼30-50 nm) is considerably larger than the size of molecular tracers (∼1-2 nm). Thus, instances of tracer (im)mobility in heterogeneous media, which provide indicators for underlying anomalous-transport mechanisms, remain obscured within the realms of SMT. Since the translation of passive tracers in an isotropic media is associated with fast dipolar rotation, we propose that authentic pauses within the LE can be revealed by probing the hindrance of SM reorientational dynamics. Here, we demonstrate how polarization-resolved SMT (PR-SMT) can provide emission anisotropy at each super-localized position, thereby revealing the tumbling propensity of SMs during random walks. For rhodamine 6G tracers undergoing heterogeneous transport in a hydrated polyvinylpyrrolidone (PVP) network, analysis of PR-SMT trajectories enabled us to discern instances of genuine immobility and localized motion within the LE. Our investigations on 100 SMs in (plasticized) PVP films reveal a wide distribution of dwell times and pause frequencies, demonstrating that most probes intermittently experience complete translational and rotational immobilization. This indicates that tracers serendipitously encounter compact, rigid polymer cavities during transport, implying the existence of nanoscale glass-like domains sparsely distributed in a predominantly deep-rubbery polymer network far above the glass transition.

Heterogeneity during Plasticization of Poly(vinylpyrrolidone): Insights from Reorientational Mobility of Single Fluorescent Probes.

While dynamics of single-molecule (SM) fluorescent probes have been used to investigate the structure and relaxation processes in polymers near the glass transition temperature (Tg), it is difficult to perform SM imaging at elevated temperatures which restricts such studies to a limited number of polymers for which Tg is close to room temperature (RT). Plasticization, solvent (or additive) induced lowering of Tg, offers an alternate avenue to access various effective temperatures in the glassy and rubbery phases of polymers under ambient conditions. By investigation of the reorientational propensity of individual Rhodamine 6G (Rh6G) probes, which is governed by rigidity/dynamics of the polymer cavities, we have explored the extent of spatiotemporal heterogeneity during moisture induced plasticization of poly(vinylpyrrolidone) (PVP), far below and near (below and above) bulk Tg. Lack of any probe reorientation suggests that the matrix remains extremely rigid up to a certain level of hydration, as expected for probes buried deep within the glassy state. At intermediate levels of hydration, SMs undergo a wide variety of rotational dynamics ranging from being static/wobbling motion to slow, hindered large-angle reorientation, as well as facile, intermittently hindered fast rotation, which reflects that swelling/softening of network cavities is spatiotemporally extremely diverse as the effective Tg approaches RT. SM probes exhibit temporally nonuniform rotational mobility even at relatively high moisture contents of the matrix beyond which probes can undergo translational motion, which indicates that relatively slow time scale polymer segmental motion can be operational for plasticized PVP (in the rubbery state). Our inferences are supported by the non-Gaussian nature of angular jump distributions for dipolar reorientation, similar to those reported for translational diffusion of SM tracers in polymers and cellular media, suggesting the existence of slow time-varying local environmental changes around individual probe molecules during plasticization.

Probing differential hydration of poly(vinylpyrrolidone) thin films using tracer mobility: an insight from fluorescence correlation spectroscopy.

Dynamics of small probe molecules have been routinely used to unravel the intrinsic details of charged ion transport in polymer brushes and polyelectrolyte multilayer (PEM) thin films. However, corresponding morphological properties affected with absorption of moisture have been hardly dealt with despite numerous applications of isotropic thin films in material chemistry and medical purposes. We have explored the overall structural changes associated with plasticization of PVP thin films by probing dynamics of small reporter (rhodamine 6G, Rh6G) molecules using fluorescence correlation spectroscopy (FCS). It was observed that under lesser amounts of absorbed moisture, the rigidity of the film matrix was high enough to inhibit appreciable molecular mobility. Nonetheless, with gradual increase in the moisture level within the film, molecular movement became extremely facile, so much so that it almost attained close to a solution like state. Molecular mobility was found to be dependent on both the method of preparation and the thickness of the thin films. The diffusivities mostly followed anomalous subdiffusive behaviors, reminiscent of dynamics of tracers in crowded cellular environments. The mobility was found to be independent of any electrostatic interaction between probe and polymer thin film. Hence, the tracer dynamics was attributed most likely to the viscoelasticity of the thin film matrix.

Plasticization of poly(vinylpyrrolidone) thin films under ambient humidity: insight from single-molecule tracer diffusion dynamics.

Studies on diffusion dynamics of single molecules (SMs) have been useful in revealing inhomogeneity of polymer thin films near and above the glass-transition temperature (T(g)). However, despite several applications of polymer thin films where exposure to solvent (or vapor) is common, the effect of absorbed solvent molecules on local morphology and rigidity of polymer matrices is yet to be explored in detail. High-T(g) hydrophilic polymers such as poly(vinylpyrrolidone) (PVP) are used as pharmaceutical coatings for drug release in aqueous medium, as they readily absorb moisture, which results in effective lowering of the T(g) and thereby leads to plasticization. The effect of moisture absorption on swelling and softening of PVP thin films was investigated by visualizing the diffusion dynamics of rhodamine 6G (Rh6G) tracer molecules at various ambient relative humidities (RH). Wide-field epifluorescence microscopy, in conjunction with high-resolution SM tracking, was used to monitor the spatiotemporal evolution of individual tracers under varied moisture contents of the matrix. In the absence of atmospheric moisture, Rh6G molecules in dry PVP films are translationally inactive, suggestive of rigid local environments. Under low moisture contents (RH 30-50%), translational mobility remains arrested but rotational motion is augmented, indicating slight swelling of the polymer network which marks the onset of plasticization. The translational mobility of Rh6G was found to be triggered only at a threshold ambient RH, beyond which a large proportion of tracers exhibit extensive diffusion dynamics. Interestingly, SM tracking data at higher moisture contents of the film (RH ≥ 60%) reveal that the distributions of dynamic parameters (such as diffusivity) are remarkably broad, spanning several orders of magnitude. Furthermore, Rh6G molecules display a wide variety of translational motion even at a fixed ambient RH, clearly pointing out the extremely inhomogeneous environment of plasticized PVP network. Intriguingly, it is observed that a majority of tracers undergo anomalous subdiffusion even under high moisture contents of the matrix. Analyses of SM trajectories using velocity autocorrelation function reveal that subdiffusive behaviors of Rh6G are likely to originate from fractional Brownian motion, a signature of tracer dynamics in viscoelastic medium.