Sex differences orchestrated by androgens at single-cell resolution.

Sex differences in mammalian complex traits are prevalent and are intimately associated with androgens1-7. However, a molecular and cellular profile of sex differences and their modulation by androgens is still lacking. Here we constructed a high-dimensional single-cell transcriptomic atlas comprising over 2.3 million cells from 17 tissues in Mus musculus and explored the effects of sex and androgens on the molecular programs and cellular populations. In particular, we found that sex-biased immune gene expression and immune cell populations, such as group 2 innate lymphoid cells, were modulated by androgens. Integration with the UK Biobank dataset revealed potential cellular targets and risk gene enrichment in antigen presentation for sex-biased diseases. This study lays the groundwork for understanding the sex differences orchestrated by androgens and provides important evidence for targeting the androgen pathway as a broad therapeutic strategy for sex-biased diseases.

Worry about COVID-19 as a predictor of future insomnia.

The coronavirus disease 2019 (COVID-19) pandemic resulted in significant increases in insomnia, with up to 60% of people reporting increased insomnia. However, it is unclear whether exposure to risk factors for the virus or worries about COVID-19 are more strongly associated with insomnia. Using a three-part survey over the course of the first 6 months of the pandemic, we evaluated associations between COVID-19 exposures, COVID-19 worries, and insomnia. We hypothesised that COVID-19-related worries and exposure to risk of COVID-19 would predict increases in insomnia. Participants (N = 3,560) completed a survey at three time-points indicating their exposures to COVID-19 risk factors, COVID-19-related worries, and insomnia. COVID-19 worry variables were consistently associated with greater insomnia severity, whereas COVID-19 exposure variables were not. COVID-19 worries decreased significantly over time, and there were significant interactions between change in COVID-19 worries and change in insomnia severity over time. Individuals who experienced increases in COVID-19 worries also experienced increases in insomnia severity. Changes in worry during the COVID-19 pandemic were associated with changes in insomnia; worries about COVID-19 were a more consistent predictor of insomnia than COVID-19 exposures. Evidence-based treatments targeting virus-related worries may improve insomnia during this and future calamities.

Factors Associated with Caregivers' Decisions to Pursue a Diagnostic Evaluation After a Positive Autism Screen in Primary Care.

PURPOSE: We sought to understand whether a child's sex, age, race, ethnicity, caregiver education, family income, and/or number of endorsed autism signs are associated with a caregiver's decision to pursue an autism diagnostic evaluation after their child received a positive autism screen. METHODS: 129 children, 17-30 months, received a positive autism screen on the Modified Checklist for Autism in Toddlers-Revised with Follow-Up, and all caregivers were offered ready access to a diagnostic evaluation by a trained professional in English or Spanish at no cost. RESULTS: 88 children received an evaluation and 41 did not. The likelihood of receiving an evaluation was associated with the child's race. Only 58.1% of Black children were evaluated, compared to 80% of Hispanic/Latino and 88.5% of White children. Children of Spanish-speaking caregivers showed high rates of evaluation completion (85.7%). Children who were evaluated versus were not evaluated did not significantly differ in terms of child's sex, number of autism signs endorsed by the caregiver, caregiver's education and preferred language (English versus Spanish), or household income. CONCLUSION: Even though the present study removed many common barriers to receiving a timely diagnostic evaluation, caregivers of Black children were less likely to pursue an autism diagnostic evaluation for their child. Future research is needed to understand the needs and perspectives of Black families to promote engagement in clinical care and reduce disparities in receiving a timely autism diagnosis which is important for accessing supports and services that can improve children's outcomes.

Exploring phenotypic overlap across schizotypy and autism spectrum conditions in American and Chinese young adults.

Competing theories have been proposed to explain the considerable overlap in social-cognitive features and risk factors across schizotypy and autism spectrum conditions (ASCs). Six previous factor analyses have been reported in the literature, yet all have major limitations; evidence for the clear superiority of any of the competing theories is insufficient and warrants further investigation. The primary aim of the present research was to identify dimensions that cut across schizotypy and ASCs while addressing limitations of past research. Data were collected from three independent samples (n = 1006, 544, and 2469) in the U.S. and China using the Autism-Spectrum Quotient, the Schizotypal Personality Questionnaire, and the Wisconsin Schizotypy Scales. Exploratory factor analyses in Sample 1 identified an interpretable three-factor structure, which was replicated in Samples 2 and 3 using confirmatory factor analyses. We found consistent evidence for three dimensions (Aberrant Salience, Asociality, and Concrete Thinking) underlying schizotypy and ASCs. This three-dimension model is consistent with a common vulnerability model of schizotypy and ASCs. Implications of these findings for the schizotypy and ASCs literature are discussed.

Sleep disorder symptoms and suicidal urges among US Marines seeking suicide treatment: Findings from an intensive daily assessment study.

BACKGROUND: Rates of suicide in United States Marines are among the highest in the military, and sleep disorder symptoms are a known risk factor for suicide in the military. Intensive ecological momentary assessments (EMA) might improve the ability to detect periods that are characterized by increased suicidal ideation. Marines who were at high risk for suicide were intensively assessed for one month on sleep, suicidal urges, posttraumatic stress disorder (PTSD) and depression symptoms. METHODS: U.S. Marines (N = 40) who had a past month suicide attempt or suicidal urges with intent were sent EMA for 28 days. Mixed effects models explored associations among daily sleep, suicidal urges, PTSD, and depression symptoms. RESULTS: Worsened sleep indicators on a given night significantly predicted higher maximum values of suicide urges the following day. Worse sleep quality the prior night was moderately associated with more severe PTSD symptoms and depression symptoms. Greater severity of PTSD symptoms and depression symptoms were strongly associated with both the maximum value and the range of suicide urges. PTSD and depression symptoms mediated the relationship between sleep quality and suicidal urges. Participants reported that 0000-0300 had the greatest elevation in endorsement of highest suicide urges. LIMITATIONS: This study had a small sample size may not generalize beyond active duty Marines. CONCLUSIONS: Poor sleep quality and other sleep markers were an important risk factor for suicidal urges among U.S. Marines. This relationship was mediated by exacerbations in PTSD and depression symptoms. Interventions are needed to interrupt suicide risk during and following nights with poor sleep.

Racial and Ethnic Disparities in Barriers to Mental Health Treatment Among U.S. College Students.

OBJECTIVE: Using a sample of U.S. college students, the authors evaluated whether barriers to mental health treatment varied by race and ethnicity. METHODS: Data were drawn from a large multicampus study conducted across 26 U.S. colleges and universities. The sample (N=5,841) included students who screened positive for at least one mental disorder and who were not currently receiving psychotherapy. RESULTS: The most prevalent barriers to treatment across the sample were a preference to deal with issues on one's own, lack of time, and financial difficulties. Black and Hispanic/Latine students reported a greater willingness to seek treatment than did White students. However, Black and Hispanic/Latine students faced more financial barriers to treatment, and Hispanic/Latine students also reported lower perceived importance of mental health. Asian American students also reported financial barriers and preferred to handle their issues on their own or with support from family or friends and had lower readiness, willingness, and intentionality to seek help than did White students. CONCLUSIONS: Disparities in unmet treatment needs may arise from both distinct and common barriers and point to the potential benefits of tailored interventions to address the specific needs of students of color from various racial and ethnic backgrounds. The findings further underscore the pressing need for low-cost and brief treatment models that can be used or accessed independently to address the most prevalent barriers for students.

Integrated characterization of hepatobiliary tumor organoids provides a potential landscape of pharmacogenomic interactions.

Despite considerable efforts to identify human liver cancer genomic alterations that might unveil druggable targets, the systematic translation of multiomics data remains challenging. Here, we report success in long-term culture of 64 patient-derived hepatobiliary tumor organoids (PDHOs) from a Chinese population. A divergent response to 265 metabolism- and epigenetics-related chemicals and 36 anti-cancer drugs is observed. Integration of the whole genome, transcriptome, chromatin accessibility profiles, and drug sensitivity results of 64 clinically relevant drugs defines over 32,000 genome-drug interactions. RUNX1 promoter mutation is associated with an increase in chromatin accessibility and a concomitant gene expression increase, promoting a cluster of drugs preferentially sensitive in hepatobiliary tumors. These results not only provide an annotated PDHO biobank of human liver cancer but also suggest a systematic approach for obtaining a comprehensive understanding of the gene-regulatory network of liver cancer, advancing the applications of potential personalized medicine.

Prevalence and nature of prior developmental and medical concerns in toddlers who screen positive for autism in primary care.

LAY ABSTRACT: The American Academy of Pediatrics recommends that all children be screened for autism at their 18- and 24-month well-child visit. For children who screen positive for autism, it is unknown whether this usually represents the first time a developmental concern has been raised or if other developmental concerns typically precede a positive autism screen. Such knowledge could help guide providers in how to appropriately convey feedback regarding autism screening. This study found that, for close to 80% of children with a positive autism screen, caregivers or providers had a prior autism, language, motor, or other developmental concern documented in the electronic health record. Many also had other prior concerns frequently linked to autism, such as sleep and gastrointestinal problems, and received physical or speech therapy. On average, prior to screening children who received a positive Modified-Checklist for Autism in Toddlers had two documented concerns by at 1 year of age and three concerns by 2 years of age. These findings imply that screening for autism as a part of routine pediatric care likely takes place in the context of larger conversations regarding existing developmental concerns, allowing for a less stigmatizing discussion of autism. Framing the presence of prior concerns in the setting of a positive screen in this context may create a reaffirming space for existing caregiver concerns and a lessened emotional burden on caregivers.

A transdiagnostic evaluation of contrast avoidance across generalized anxiety disorder, major depressive disorder, and social anxiety disorder.

BACKGROUND: The contrast avoidance model (CAM) proposes that persons with generalized anxiety disorder (GAD) are sensitive to sharp increases in negative emotion or decreases in positive emotion (i.e., negative emotional contrasts; NEC) and use worry to avoid NEC. Sensitivity to and avoidance of NEC could also be a shared feature of major depressive disorder (MDD) and social anxiety disorder (SAD). METHODS: In a large college sample (N = 1409), we used receiver operating characteristics analysis to examine the accuracy of a measure of emotional contrast avoidance in detecting probable GAD, MDD, and SAD. RESULTS: Participants with probable GAD, MDD, and SAD all reported higher levels of contrast avoidance than participants without the disorder (Cohen's d = 1.32, 1.62 and 1.53, respectively). Area under the curve, a measure of predictive accuracy, was 0.81, 0.87, and 0.83 for predicting probable GAD, MDD, and SAD, respectively. A cutoff score of 48.5 optimized predictive accuracy for probable GAD and SAD, and 50.5 optimized accuracy for probable MDD. CONCLUSION: A measure of emotional contrast avoidance demonstrated excellent ability to predict probable GAD, MDD, and SAD. Sensitivity to and avoidance of NEC appears to be a transdiagnostic feature of these disorders.

Metabolic classification suggests the GLUT1/ALDOB/G6PD axis as a therapeutic target in chemotherapy-resistant pancreatic cancer.

Metabolic reprogramming is known as an emerging mechanism of chemotherapy resistance, but the metabolic signatures of pancreatic ductal adenocarcinomas (PDACs) remain unclear. Here, we characterize the metabolomic profile of PDAC organoids and classify them into glucomet-PDAC (high glucose metabolism levels) and lipomet-PDAC (high lipid metabolism levels). Glucomet-PDACs are more resistant to chemotherapy than lipomet-PDACs, and patients with glucomet-PDAC have a worse prognosis. Integrated analyses reveal that the GLUT1/aldolase B (ALDOB)/glucose-6-phosphate dehydrogenase (G6PD) axis induces chemotherapy resistance by remodeling glucose metabolism in glucomet-PDAC. Increased glycolytic flux, G6PD activity, and pyrimidine biosynthesis are identified in glucomet-PDAC with high GLUT1 and low ALDOB expression, and these phenotypes could be reversed by inhibiting GLUT1 expression or by increasing ALDOB expression. Pharmacological inhibition of GLUT1 or G6PD enhances the chemotherapy response of glucomet-PDAC. Our findings uncover potential metabolic heterogeneity related to differences in chemotherapy sensitivity in PDAC and develop a promising pharmacological strategy for patients with chemotherapy-resistant glucomet-PDAC through the combination of chemotherapy and GLUT1/ALDOB/G6PD axis inhibitors.

Predictive Value of Early Autism Detection Models Based on Electronic Health Record Data Collected Before Age 1 Year.

Importance: Autism detection early in childhood is critical to ensure that autistic children and their families have access to early behavioral support. Early correlates of autism documented in electronic health records (EHRs) during routine care could allow passive, predictive model-based monitoring to improve the accuracy of early detection. Objective: To quantify the predictive value of early autism detection models based on EHR data collected before age 1 year. Design, Setting, and Participants: This retrospective diagnostic study used EHR data from children seen within the Duke University Health System before age 30 days between January 2006 and December 2020. These data were used to train and evaluate L2-regularized Cox proportional hazards models predicting later autism diagnosis based on data collected from birth up to the time of prediction (ages 30-360 days). Statistical analyses were performed between August 1, 2020, and April 1, 2022. Main Outcomes and Measures: Prediction performance was quantified in terms of sensitivity, specificity, and positive predictive value (PPV) at clinically relevant model operating thresholds. Results: Data from 45 080 children, including 924 (1.5%) meeting autism criteria, were included in this study. Model-based autism detection at age 30 days achieved 45.5% sensitivity and 23.0% PPV at 90.0% specificity. Detection by age 360 days achieved 59.8% sensitivity and 17.6% PPV at 81.5% specificity and 38.8% sensitivity and 31.0% PPV at 94.3% specificity. Conclusions and Relevance: In this diagnostic study of an autism screening test, EHR-based autism detection achieved clinically meaningful accuracy by age 30 days, improving by age 1 year. This automated approach could be integrated with caregiver surveys to improve the accuracy of early autism screening.

COVID-19 Worries and Insomnia: A Follow-Up Study.

The COVID-19 pandemic was associated with significant increases in sleep disorder symptoms and chronic worry. We previously demonstrated that worry about the pandemic was more strongly associated with subsequent insomnia than the converse during the acute (first 6 months) phase of the pandemic. In this report, we evaluated whether that association held over one year of the pandemic. Participants (n = 3560) completed self-reported surveys of worries about the pandemic, exposure to virus risk factors, and the Insomnia Severity Index on five occasions throughout the course of one year. In cross-sectional analyses, insomnia was more consistently associated with worries about the pandemic than exposure to COVID-19 risk factors. In mixed-effects models, changes in worries predicted changes in insomnia and vice versa. This bidirectional relationship was further confirmed in cross-lagged panel models. Clinically, these findings suggest that during a global disaster, patients who report elevations in either worry or insomnia should be considered for evidence-based treatments for these symptoms to prevent secondary symptoms in the future. Future research should evaluate the extent to which dissemination of evidence-based practices for chronic worry (a core feature of generalized anxiety disorder or illness anxiety disorder) or insomnia reduces the development of co-occurring symptoms during a global disaster.

A Suicide Prevention Intervention for Emerging Adult Sexual and Gender Minority Groups: Protocol for a Pilot Hybrid Effectiveness Randomized Controlled Trial.

BACKGROUND: Suicide attempts and suicide death disproportionately affect sexual and gender minority emerging adults (age 18-24 years). However, suicide prevention strategies tailored for emerging adult sexual and gender minority (EA-SGM) groups are not widely available. The Safety Planning Intervention (SPI) has strong evidence for reducing the risk for suicide in the general population, but it is unclear how best to support EA-SGM groups in their use of a safety plan. Our intervention (Supporting Transitions to Adulthood and Reducing Suicide [STARS]) builds on content from an existing life skills mobile app for adolescent men who have sex with men (iREACH) and seeks to target core risk factors for suicide among EA-SGM groups, namely, positive affect, discrimination, and social disconnection. The mobile app is delivered to participants randomized to STARS alongside 6 peer mentoring sessions to support the use of the safety plan and other life skills from the app to ultimately reduce suicide risk. OBJECTIVE: We will pilot-test the combination of peer mentoring alongside an app-based intervention (STARS) designed to reduce suicidal ideation and behaviors. STARS will include suicide prevention content and will target positive affect, discrimination, and social support. After an in-person SPI with a clinician, STARS users can access content and activities to increase their intention to use SPI and overcome obstacles to its use. EA-SGM groups will be randomized to receive either SPI alone or STARS and will be assessed for 6 months. METHODS: Guided by the RE-AIM (reach, efficacy, adoption, implementation, and maintenance) framework, we will recruit and enroll a racially and ethnically diverse sample of 60 EA-SGM individuals reporting past-month suicidal ideation. Using a type-1 effectiveness-implementation hybrid design, participants will be randomized to receive SPI (control arm) or to receive SPI alongside STARS (intervention arm). We will follow the participants for 6 months, with evaluations at 2, 4, and 6 months. Preliminary effectiveness outcomes (suicidal ideation and behavior) and hypothesized mechanisms of change (positive affect, coping with discrimination, and social support) will serve as our primary outcomes. Secondary outcomes include key implementation indicators, including participants' willingness and adoption of SPI and STARS and staff's experiences with delivering the program. RESULTS: Study activities began in September 2021 and are ongoing. The study was approved by the institutional review board of the University of Pennsylvania (protocol number 849500). Study recruitment began on October 14, 2022. CONCLUSIONS: This project will be among the first tailored, mobile-based interventions for EA-SGM groups at risk for suicide. This project is responsive to the documented gaps for this population: approaches that address chosen family, focus on a life-course perspective, web approaches, and focus on health equity and provision of additional services relevant to sexual and gender minority youth. TRIAL REGISTRATION: ClinicalTrials.gov NCT05018143; https://classic.clinicaltrials.gov/ct2/show/NCT05018143. INTERNATIONAL REGISTERED REPORT IDENTIFIER (IRRID): DERR1-10.2196/48177.

Phenotypic predictors of suicide subtypes from pre-to postdeployment in active duty military personnel.

Military service members are at increased risk for suicide, but there are few strategies for detecting those who are at highest risk after a deployment. Using all available data collected from 4119 Military service members before and after their deployment to Iraq for Operation Iraqi Freedom, we tested whether predeployment characteristics clustered together to predict postdeployment suicidal risk. Latent class analysis showed that three classes best characterized the sample at predeployment. Class 1 had significantly higher scores on PTSD severity pre- and postdeployment than Classes 2 and 3 (Ps < .001). At postdeployment, Class 1 also had a greater proportion of endorsement of lifetime and past year suicidal ideation than Classes 2 and 3 (Ps < .05) and a greater proportion of lifetime suicide attempts than Class 3 (P < .001). Class 1 also had a greater proportion of endorsement of past-30-days intention to act on suicidal thoughts than Classes 2 and 3 (Ps < .05) and past-30-days specific plan for suicide than Classes 2 and 3 (Ps < .05). The study showed that based only on predeployment data, it is possible to determine which service members might be at highest risk for suicidal ideation and behavior at postdeployment.

Idiographic Dynamics between Suicide Ideation and Depression, Anxiety and Posttraumatic Stress Symptoms in Persons Living with HIV: A Pilot Study.

BACKGROUND: Given that suicide ideation (SI) fluctuates drastically over short periods of time and is heterogenous across individuals, idiographic suicide research is warranted. In this pilot study, we used intensive ecological momentary assessment (EMA) to examine whether anxiety, depression, and PTSD symptoms on a given day predicted next-day SI on a person-to-person basis. METHODS: PLWH (N = 10) with past-month SI completed daily randomly assessed ratings of suicidal urges using the Suicide-Visual Analogue Scale (S-VAS) and daily assessed ratings of anxiety, depression, and PTSD symptoms for 28 days. We used N = 1 Dynamic Structural Equation Modeling to test whether depression, anxiety or PTSD symptoms in the prior day predicted next-day S-VAS for each individual. RESULTS: Across all participants, S-VAS on a given day was not predicted by prior-day anxiety, PTSD symptoms or S-VAS. In one participant, higher depression symptoms predicted lower next-day S-VAS. CONCLUSIONS: Daily-level data may be insufficient to predict near-term increases in suicide risk based on anxiety, depression, or PTSD symptoms in PLWH. These findings suggest the importance of finer-grained assessments (e.g., assessing suicide risk and its correlates multiple times per day) to better understand changes in suicide risk over time among PLWH.

Testing a theory-driven factor structure of the autism-spectrum quotient.

The Autism-Spectrum Quotient (AQ) is a popular instrument used to assess the degree to which individuals exhibit features of autism spectrum conditions (ASC). The current study aimed to develop a theory-driven factor structure of the AQ that would fit as well across samples as the 12 previously proposed factor structures, all of which, except for the original Baron-Cohen model, had been developed on the basis of exploratory factor analysis (EFA) or principal component analysis. We first proposed a six-factor solution: (1) social anhedonia; (2) interest in details/patterns; (3) imagination ability; (4) desire for predictability/routine; (5) social cognition; and (6) social discourse convention. We tested the six-factor structure and made final item selections (27 items) with EFA using data from college students (n = 503). Then, we empirically tested alternative factor structure models in three other independent samples (ns = 503; 1263; 1641) using confirmatory factor analysis. Results indicated that our model fit as well, if not better, than all of the other models across samples, regardless of parameter estimation methods and software packages. Overall, the theory-driven replicable six-factor structure that we report holds the potential to be used to measure the six domains of features that we identified in the AQ. LAY SUMMARY: Questionnaire measures of autism spectrum conditions have typically been used to measure approximately four broad dimensions. Our study suggests that the Autism-Spectrum Quotient can be used to measure six more narrowly defined dimensions: social anhedonia, interest in details/patterns, imagination ability, desire for predictability/routine, social cognition, and social discourse convention. Additional work is needed to develop measures of a much wider variety of autism spectrum features.

Social anxiety does not impair attention inhibition: An emotion anti-saccade task.

BACKGROUND AND OBJECTIVES: Attention avoidance and attention vigilance are two typical attentional biases in individuals with social anxiety disorder (SAD). Attention inhibition is a type of attention control, which may be the key factor affecting attention vigilance and attention avoidance. However, previous studies have not examined the difference between the attention inhibition in individuals with SAD and healthy controls. METHODS: To further explore this question, the current study used the single anti-saccade task with emotional facial stimuli to assess attention inhibition in 27 individuals with SAD and 22 healthy controls. RESULTS: Regardless of the emotional valence of the facial stimuli, error rates in the social anxiety group were lower than that of the healthy control group, but there was no significant group difference in the saccade latency. LIMITATIONS: This research only examined the attentional inhibition process highly related to attention avoidance and attention vigilance. Future research may benefit from adopting different research paradigms for more robust and generalizable conclusions. CONCLUSIONS: Results suggest that individuals with SAD have better attention inhibition abilities than healthy control. Such enhanced attention inhibition may underlie their avoidance of threatening social cues.

The unique contribution of depression to cognitive impairment in Post-Acute Sequelae of SARS-CoV-2 infection.

Background: Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection is associated with significant cognitive impairment and increased risk for mental health comorbidities. This study aimed to identify specific associations between cognitive impairment, self-reported disruptions in memory, and psychiatric symptoms including depression, anxiety, posttraumatic stress disorder (PTSD), and perceived sleep concerns. Methods: Data collected from all consecutive patients with Post-Acute Sequelae of SARS-CoV-2 infection (PASC) who presented to a dedicated Post-COVID Clinic were used to evaluate whether certain psychiatric symptoms were more strongly associated with cognitive impairment and self-reported memory disturbances. Results: Univariate and multivariable analyses revealed that depression symptom severity was significantly associated with the severity of cognitive impairment among patients with PASC. This association was driven primarily by lower performance on verbal fluency, attention, and delayed recall tasks among patients with higher depression symptoms severity. Perceived sleep concerns were an important predictor of self-reported memory disturbances. Conversely, neither PTSD symptom severity nor anxiety symptom severity were significant predictors of cognitive impairment or self-reported memory disturbances. Conclusions: These findings have important clinical implications for justifying the need for screening patients with PASC for both depression and cognitive impairment.

Deep Learning-Based Prediction of Hematoma Expansion Using a Single Brain Computed Tomographic Slice in Patients With Spontaneous Intracerebral Hemorrhages.

OBJECTIVES: We aimed to predict hematoma expansion in intracerebral hemorrhage (ICH) patients by using the deep learning technique. METHODS: We retrospectively collected data from ICH patients treated between May 2015 and May 2019. Head computed tomography (CT) scans were performed at admission, and 6 hours, 24 hours, and 72 hours after admission. CT scans were mandatory when neurologic deficits occurred. Univariate and multivariate analyses were conducted to illustrate the association between clinical variables and hematoma expansion. Convolutional neural network (CNN) was adopted to predict hematoma expansion based on brain CT slices. In addition, 5 machine learning methods, including support vector machine, multi-layer perceptron, naive Bayes, decision tree, and random forest, were also performed to predict hematoma expansion based on clinical variables for comparisons. RESULTS: A total of 223 patients were included. It was revealed that patients' older age (odds ratio [95% confidence interval]: 1.783 [1.417-1.924]), cerebral hemorrhage and breaking into the ventricle (2.524 [1.291-1.778]), coagulopathy (2.341 [1.677-3.454]), and baseline National Institutes of Health Stroke Scale (1.545 [1.132-3.203]) and Glasgow Coma Scale scores (0.782 [0.432-0.918]) independently associated with hematoma expanding. After 4-5 epochs, the CNN framework was well trained. The average sensitivity, specificity, and accuracy of CNN prediction are 0.9197, 0.8837, and 0.9058, respectively. Compared with 5 machine learning methods based on clinical variables, CNN can also achieve better performance. CONCLUSIONS: More than 90% of hematomas with or without expansion can be precisely classified by deep learning technology within this study, which is better than other methods based on clinical variables only. Deep learning technology could favorably predict hematoma expansion from non-contrast CT scan images.