Importance:Homelessness is associated with negative health outcomes and increased health care costs. The United States Department of Veterans Affairs (VA) Supportive Services for Veteran Families (SSVF) program provides housing-related financial assistance and other supports to veterans experiencing housing instability; however, little is known regarding short-term assistance interventions with a prevention focus. Objective:To estimate potential impacts of the SSVF program in mortality and health care cost outcomes over 3 years following program entry. Design, Setting, and Participants:Using observational data, outcomes were compared between veterans who enrolled in SSVF with those who did not for each month from October 2015 to December 2018. A propensity score for SSVF enrollment was calculated using observable characteristics including demographics, housing history, health care cost history, comorbidities, and geography. Using inverse probability of treatment weighting-a propensity score-based method that creates a pseudopopulation in which treatment groups are balanced on observed covariates-the potential impacts of SSVF enrollment in mortality were estimated using a Cox proportional hazards regression and health care costs with a generalized linear model over the 3 years following the trial index date. Data were from the VA electronic health record for a cohort of veterans receiving care in the VA system. Each trial drew on veterans with evidence of homelessness in structured and unstructured medical records during the previous month. Data were analyzed from November 1, 2023, to September 9, 2025. Exposure:The exposure was enrollment in the SSVF program, from the Homeless Management Information System data. Main outcome:The main outcomes were all-cause mortality and VA health care costs. Results:The cohort consisted of 693 383 patient-trials with 26 649 (3.8%) enrolling in SSVF (mean [SD] age, 52.7 [12.6] years; 89.6% male) and 666 734 (96.5%) in the no SSVF group (mean [SD] age, 53.8 [13.0] years; 90.8% male). Enrollment in SSVF was associated with a decrease in the risk of mortality (hazard ratio, 0.87; 95% CI, 0.82-0.92). In addition, enrollment in SSVF was associated with an increase in outpatient costs ($7534; 95% CI, $6767-$8302) and a decrease in inpatient costs (-$10 020; 95% CI, -$13 644 to -$6396). Conclusions and Relevance:In this study, federal prevention solutions to homelessness were associated with improved health outcomes and lower inpatient costs, which should inform national policy debates within and beyond the VA.

BackgroundSubstance use disorder (SUD) remains a major public health crisis in the United States, with significant challenges in treatment access, retention, and workforce capacity. SUD care teams, including addiction medicine physicians and peer recovery coaches (PRCs), support patients receiving SUD treatment but face heavy workloads and burnout. Artificial intelligence (AI) innovations, particularly large language model (LLM)–based chatbots, may extend PRC support and provide patients with on-demand recovery support between clinic visits and PRC contacts. However, evidence on their development, feasibility, acceptability, and usability in addiction services remains limited. ObjectiveThis study describes the development, feasibility, acceptability, and usability of an AI-powered health coaching chatbot (Suzy) designed to support patients in SUD recovery. MethodsA total of 2 clinicians, 5 researchers, and 2 technology developers led a small, multiphase pilot study. In the formative phase, they conducted focus groups and qualitative in-depth interviews with 12 health care professionals and 8 patients with substance use histories to specify chatbot functions and develop a rule-based chatbot. In phase 2, they conducted usability testing of the rule-based chatbot with 8 patients who reported substance use and completed standardized tasks, surveys, and qualitative interviews. Measures included the System Usability Scale (SUS), Net Promoter Score (NPS), and Single Ease of Use Question (SEQ). In phase 3, they developed an LLM-based chatbot co-designed and fine-tuned with PRCs and other SUD experts. ResultsRule-based chatbot functions included craving management, appointment reminders, resource referrals, care team contacts, and goal setting. Usability task testing supported feasibility. In this small pilot sample, quantitative and qualitative feedback indicated acceptability and usability, with an average SUS score of 93 (benchmark 68), an NPS of 63 (benchmark 35), and a mean SEQ score of 6.5/7. Patients valued Suzy’s approachable, nonjudgmental language and features that promoted accountability, self-monitoring, and 24/7 availability, while emphasizing that chatbots should supplement but not replace human support. The LLM-based chatbot development emphasized information accuracy, safety escalation protocols to mitigate risks of inappropriate chatbot responses, human-in-the-loop features, and expanded conversational flexibility and personal tailoring. ConclusionsIn this pilot study, a rule-based chatbot designed to support SUD care demonstrated feasibility, usability, and acceptability. LLM-based chatbot development required more robust safety and emergency reporting features, while offering more patient-responsive conversational functions. By providing on-demand coaching, referrals, and reminders, Suzy may extend the reach of care teams, alleviate provider burden, and enhance patient engagement. Additional work is needed to understand how to best integrate Suzy into patients’ recovery journeys to ensure human support remains accessible and prioritized. LLM evaluation was based on expert testing and safety review. Clinical effectiveness, including the impact on substance use, was not evaluated. Next steps include evaluating the LLM chatbot in real-world settings with larger samples and assessing its efficacy in reducing substance use.

Abstract Background Approximately 70% of people experiencing homelessness smoke cigarettes, compared to 9.9% in the US general population. Barriers to quitting and sustaining cessation include structural inequities (e.g., lack of housing; limited treatment access), and high rates of serious mental illness and substance use disorders. Existing cessation treatments are often brief, not integrated into community service settings, and lack ongoing behavioral support. Community pharmacies represent a promising strategy to expand access to evidence-based tobacco treatment; however, this approach has primarily been tested in short-duration interventions. We developed the Extended Intervention for Tobacco Use (EXIT), which links extended pharmacotherapy with wellness-focused telephone coaching. This protocol describes a randomized controlled trial evaluating the feasibility, acceptability, and preliminary efficacy of EXIT. Methods This two-arm, parallel-group randomized controlled trial will be conducted in transitional shelters in Los Angeles and San Francisco. We will enroll 150 adults experiencing homelessness who smoke at least five cigarettes per day (expired carbon monoxide [CO] ≥ 8 ppm) and intend to quit within six months. Participants will be randomized 1:1 to EXIT or a pharmacist-only comparator condition (“pharm-only”). The pharm-only condition includes a single pharmacist-delivered 5As telephone session and three months of nicotine replacement therapy (NRT). The EXIT intervention includes a 5As session, six months of NRT, and 14 wellness-focused telephone coaching sessions delivered over six months by trained health coaches. The primary outcome is adherence, measured by the number of coaching sessions attended and adherence to NRT. Secondary outcomes include biochemically verified 7-day point prevalence abstinence at 3 and 6 months (CO ≤ 5 ppm). Assessments occur at baseline, 1, 3, and 6 months and include tobacco use, nicotine dependence, health behaviors (e.g., diet; physical activity), mental health, substance use, and NRT adherence. Discussion This study evaluates the first known intervention that pairs extended-duration pharmacotherapy with wellness-focused telephone coaching within transitional shelters. By integrating tobacco treatment into homeless services and pharmacy infrastructure, EXIT is designed to reduce structural barriers to sustained cessation. Findings will inform a future fully powered trial to provide a scalable model for delivering tobacco cessation treatment to people experiencing homelessness. Trial registration ClinicalTrials.gov #: NCT07148232; Registration date: 2025-08-22.

BACKGROUND AND AIMS:Individuals who engage in illicit or nonmedical opioid use may have elevated risk of health and social consequences, including progression to opioid use disorder (OUD). Preventive interventions to reduce this risk are lacking. This trial tested the impact of a primary care-integrated collaborative care approach for reducing risky opioid use, defined as nonmedical use of prescription opioids or any use of illicit opioids. DESIGN:Cluster-randomized controlled trial randomized primary care providers (PCPs) and their patients into the Subthreshold Opioid Use Disorder Prevention (STOP) intervention or enhanced usual care (EUC). SETTING:Primary care clinics at 5 U.S. sites. PARTICIPANTS:PCPs and their patients were recruited January 2021-May 2023. A total of 119 PCP clusters (STOP = 48, EUC = 51) and 202 patients (STOP = 88, EUC = 114) enrolled. Eligible patients were adults (≥18 years) having current risky opioid use, without moderate-severe OUD. Patient participants were majority female (63.4%), white (70.8%) and non-Hispanic (96.5%), with a mean age of 55.7 [standard deviation (SD) = 12.7] years. At baseline, 63.4% of participants had moderate-severe pain (Brief Pain Inventory) and below average physical (79.2%) and mental (62.4%) health (SF-12). INTERVENTIONS:The STOP collaborative care intervention consisted of brief advice from the PCP about reducing risky opioid use, meetings with a clinic-embedded nurse care manager over 12 months and remote health coaching (2-6 sessions). Both groups received primary care treatment as usual and overdose risk reduction materials. MEASUREMENTS:The primary outcome was total days of risky opioid use, recorded from 6 monthly electronic surveys. A key secondary outcome was moderate-severe OUD at 6 and 12 months. FINDINGS:A total of 77 (87.5%) STOP and 107 (93.9%) EUC participants completed the 6-month assessment period. The primary outcome analysis used the Intention-to-Treat sample with multiple imputations of missing data. Mean days of risky opioid use at 180 days were lower in STOP than EUC [12.2 (SD = 27.73) vs. 15.5 (SD = 32.64)]; the difference between groups adjusted for baseline risky opioid use was not statistically significant (rate ratio 0.95, 95% confidence interval = 0.52-1.74). One STOP participant (1.1%) and 13 EUC participants (11.4%) developed moderate-severe OUD at 6 months, and 3 (3.4%) STOP and 6 (5.3%) EUC participants had moderate-severe OUD at 12 months (P < 0.001). CONCLUSIONS:This cluster-randomized controlled trial did not find evidence that the STOP intervention for reducing risky opioid use produced greater reductions over 6 months compared with enhanced usual care, though fewer intervention participants progressed to moderate-severe opioid use disorder. Patients had a high burden of pain and comorbidities that may present challenges to reducing opioid use.

BackgroundCannabis use is increasingly prevalent among adults with depression and anxiety, raising concerns about potential interactions with prescribed antidepressant medication and/or benzodiazepines. Using data from a large, university health system in Los Angeles, this cross-sectional study examined the association between having an active prescription of antidepressant medication and/or benzodiazepines and frequency of cannabis use by sex in adult primary care patients with depression and/or anxiety.MethodsThis analysis included 45,693 adult male and female patients with a current diagnosis of depression and/or anxiety (International Classification of Diseases, Tenth Revision [ICD-10] codes: F41, F33) listed in their electronic health record (EHR). Among patients with a diagnosis, those with an active prescription for antidepressant medication and/or benzodiazepines were identified using EHR data. Cannabis use was assessed as part of routine patient self-administered screening via the EHR-based Tobacco and Cannabis Questionnaire (TCQ) from January 2021 to June 2023. Sex-stratified multivariable ordinal logistic regression models assessed whether active antidepressant medication and/or benzodiazepine prescriptions were associated with increased cannabis use frequency. Models were adjusted for sexual identity, age, race/ethnicity, employment, and Charlson Comorbidity Index (CCI).ResultsFemales comprised 67.5% of the sample of patients with depression and/or anxiety; 16.3% reported cannabis use for mental health symptoms. Cannabis use for mental health symptoms among males was 21.2%.Among those with depression and/or anxiety, 46.2% of females and 42.8% of males were prescribed antidepressant medication and/or benzodiazepines. Females with an active prescription for antidepressant medication and/or benzodiazepines had 1.32 times the adjusted odds of more frequent cannabis use (95% CI: 1.24-1.40) compared to those without a prescription. Among males, the adjusted odds ratio was lower at 1.14 (95% CI: 1.06-1.23), suggesting a significant difference by sex.ConclusionsFindings indicate that cannabis use commonly occurs alongside prescribed mental health treatment, underscoring the need for primary care providers to routinely address cannabis and medication co-use during clinical care.

BackgroundA majority of the 8.9 million Americans with opioid misuse have mild or no symptoms of opioid use disorder (OUD), but they may be at elevated risk of developing more severe OUD, overdose, or other health consequences of opioid use. The "Subthreshold Opioid Use Disorder Prevention"(STOP) Trial is evaluating a collaborative care intervention for risky opioid use in primary care. Here, we describe baseline characteristics of participants to understand their needs and assess the generalizability of the sample.MethodsRecruitment at five primary care sites spanned March 2021-May 2023. Adult patients who screened positive for subthreshold OUD (current illicit or non-medical opioid use without meeting DSM-5 criteria for moderate-severe OUD) were eligible. Baseline assessments measured self-reported demographic characteristics, other substance use, pain, and physical and mental health symptoms. Descriptive statistics summarize characteristics of the enrolled sample across sites.ResultsAmong the 202 participants, the majority identified as female (63.4%), white (70.8%), and non-Hispanic (96.5%), with mean age 55.7 (SD: 12.7) years. Nearly half (49.0%) had problem or high-risk use of prescription opioids, and most received a prescription for opioid medication in the past six months (74.8%). Many participants reported current problem use or high-risk use of alcohol (47.0%) or cannabis (31.2%). Approximately one-third endorsed mental health symptoms, including moderate-severe anxiety (35.6%), depression (31.2%), or sleep disturbance (29.7%), and 20.3% reported a past suicide attempt. In the prior six months, 14.7% had experienced a nonfatal overdose. Moderate-severe pain was reported by 63.4%, and 60.4% rated their general health as fair or poor.ConclusionsPatients with subthreshold OUD had high rates of polysubstance use and comorbidities that may present challenges to reducing risky opioid use. The STOP trial presents an opportunity to detect and address subthreshold OUD in a cohort with considerable medical and social needs, within primary care settings.Clinical trials registrationClinicalTrials.gov NCT04218201

BACKGROUND:Standardized patients (SPs) prepare medical students for difficult conversations with patients. Despite their value, SP-based simulation training is constrained by available resources and competing clinical demands. Researchers are turning to artificial intelligence and large language models, such as generative pretrained transformers, to create communication training that incorporates virtual simulated patients (VSPs). GPT-4 is a large language model advance allowing developers to design virtual simulation scenarios using text-based prompts instead of relying on branching path simulations with prescripted dialogue. These nascent developmental practices have not taken root in the literature to guide other researchers in developing their own simulations. OBJECTIVE:This study aims to describe our developmental process and lessons learned for creating a GPT-4-driven VSP. We designed the VSP to help medical student learners rehearse discussing abnormal mammography results with a patient as a primary care physician (PCP). We aimed to assess GPT-4's ability to generate appropriate VSP responses to learners during spoken conversations and provide appropriate feedback on learner performance. METHODS:A research team comprised of physicians, a medical student, an educator, an SP program director, a learning experience designer, and a health care researcher conducted the study. A formative phase with in-depth knowledge user interviews informed development, followed by a development phase to create the virtual training module. The team conducted interviews with 5 medical students, 5 PCPs, and 5 breast cancer survivors. They then developed a VSP using simulation authoring software and provided the GPT-4-enabled VSP with an initial prompt consisting of a scenario description, emotional state, and expectations for learner dialogue. It was iteratively refined through an agile design process involving repeated cycles of testing, documenting issues, and revising the prompt. As an exploratory feature, the simulation used GPT-4 to provide written feedback to learners about their performance communicating with the VSP and their adherence to guidelines for difficult conversations. RESULTS:In-depth interviews helped establish the appropriate timing, mode of communication, and protocol for conversations between PCPs and patients during the breast cancer screening process. The scenario simulated a telephone call between a physician and patient to discuss the abnormal results of a diagnostic mammogram that that indicated a need for a biopsy. Preliminary testing was promising. The VSP asked sensible questions about their mammography results and responded to learner inquiries using a voice replete with appropriate emotional inflections. GPT-4 generated performance feedback that successfully identified strengths and areas for improvement using relevant quotes from the learner-VSP conversation, but it occasionally misidentified learner adherence to communication protocols. CONCLUSIONS:GPT-4 streamlined development and facilitated more dynamic, humanlike interactions between learners and the VSP compared to branching path simulations. For the next steps, we will pilot-test the VSP with medical students to evaluate its feasibility and acceptability.

Depression, anxiety, and cannabis use are growing, interconnected primary care concerns, but remain understudied due few health systems conducting systematic cannabis use screening. This study examines cannabis use and risk of cannabis use disorder (CUD) among primary care patients, comparing outcomes by depression and anxiety diagnoses and psychotropic prescriptions. We assessed past three-month cannabis use, reasons for use, and risk of CUD among 170,032 adult primary care patients at a large health system in Los Angeles, CA under a routine screening protocol using the validated, self-administered ASSIST survey. This survey was embedded in the electronic health record (EHR), where data on ICD-10 diagnostic codes for depressive and anxiety disorders, psychotropic prescriptions, and demographics were collected. Logistic regression analysis assessed the association of depression and anxiety diagnoses on risk of CUD. Median age was 48 years (IQR 35–61), 57.8

This study used electronic health record (EHR) data from 9869 sexual minority patients aged 18 and older who had a primary care visit within a Los Angeles university health system (June 2020-May 2023). Sexual minority patients, defined as lesbian, gay, or bisexual, were screened for cannabis use and risk of cannabis use disorder (CUD) during annual wellness visits. Mental health diagnoses were extracted from patients' EHR. Multivariable regression models were used to assess associations between sexual identity, sex, and cannabis-related outcomes, including cannabis use, risk of CUD, mental health diagnoses among individuals reporting cannabis use, and cannabis use for symptom management among those with a corresponding diagnosis. All models controlled for age and race/ethnicity. Bisexual patients had higher adjusted odds of cannabis use compared to gay/lesbian patients (AOR females: 1.67; males: 1.47). Bisexual males had greater odds of risk of CUD (AOR: 1.48) and depression diagnosis (AOR:1.86) compared to gay males. Bisexual males with a diagnosis for depression had higher odds of cannabis use for managing depression or sadness symptoms (AOR: 2.44) compared to gay males. Bisexual females had higher odds of a severe stress diagnosis compared to gay/lesbian females (AOR: 2.44). These findings emphasize the importance of developing targeted primary care approaches that address both cannabis use and mental health concerns among bisexual patients. PURPOSE:While the prevalence of cannabis use is higher among sexual minorities as compared to their heterosexual counterparts, few studies have examined the differences in cannabis use across sex and sexual identity among sexual minority individuals, especially in the context of health care. This study examines the association of sexual identity and sex with cannabis use, risk of cannabis use disorder (CUD), symptoms managed with cannabis use, and mental health diagnoses among sexual minority primary care patients. METHODS:We conducted a cross-sectional study using electronic health record (EHR) data from 9869 patients ≥ 18 years of age who identified as sexual minority, defined as lesbian, gay, or bisexual, and had a primary care visit between June 2020 and May 2023 within a university-based health system in Los Angeles, CA. Routine screening for past 3-month cannabis use and risk of CUD was based on the Alcohol Substance Involvement Screening Test (ASSIST) and was conducted as part of all annual wellness visits. Patients were asked to report symptoms for which they used cannabis, and mental health diagnoses were extracted from patients' EHR. Diagnoses meeting clinical threshold criteria were identified based on the International Classification of Diseases, Tenth Revision (ICD-10) codes and included anxiety disorders (ICD-10 F41), depressive disorders (ICD-10 F33), and severe stress (ICD-10 F43). Differences in the prevalence of cannabis use by sociodemographic characteristics were stratified by sex and compared across sexual identity using chi-squared tests for categorical variables and Wilcoxon rank-sum tests for continuous variables. Multivariable regression models were used to assess associations between sexual identity, sex, and cannabis-related outcomes, including cannabis use in the past 3 months, risk of CUD, mental health diagnoses among individuals reporting cannabis use, and cannabis use for symptom management among those with a corresponding diagnosis. All models controlled for age and race/ethnicity. RESULTS:Among the 9869 patients included in this study, 30.7 % reported cannabis use in the past 3 months. Bisexual patients had higher adjusted odds of cannabis use in comparison to gay/lesbian patients in both females and males (female adjusted odds ratio (AOR):1.67; 95 % Confidence Interval (CI) 1.46, 1.92; male AOR: 1.47; 95 % CI 1.24, 1.73). Among males reporting cannabis use, bisexual individuals had greater odds of risk of CUD (AOR: 1.48; 95 % CI: 1.14, 1.93) and depression diagnosis (AOR: 1.86; 95 % CI: 1.34, 2.56) compared to gay males. Bisexual males with a diagnosis for depression had higher odds of cannabis use for managing depression or sadness symptoms (AOR: 2.44; 95 % CI: 1.34, 4.52) compared to gay males. Among females reporting cannabis use, bisexual patients had higher odds of a severe stress diagnosis compared to gay/lesbian females (AOR: 2.44; 95 % CI: 1.45, 4.35). CONCLUSIONS:These findings highlight the need for primary care providers to consider the unique experiences of bisexual patients, particularly regarding mental health and cannabis use. Increased odds of cannabis use among bisexual patients, coupled with their higher odds of mental health disorders, reinforces the importance of integrating mental health support and addressing cannabis use in routine healthcare.

Background Homelessness is a growing concern in the United States, especially among people who use drugs (PWUD). The degree of material hardship among this population may be linked to worse health outcomes. PWUD experiencing homelessness in urban areas are increasingly subjected to policies and social treatment, such as forced displacement, which may worsen material hardship. It is critical to describe hardship among PWUD and examine if it is linked to health outcomes. Methods Data were collected as part of a prospective cohort study of PWUD in Los Angeles, California and Denver, Colorado (n = 476). Analysis sample size was smaller (N = 395) after selecting for people experiencing homelessness and for whom data were complete. Five indicators assessing hardship (difficulty finding food, clothing, restrooms, places to wash/shower, and shelter) in the past three months were obtained from participants at baseline and were used in latent class analysis (LCA). We chose a base latent class model after examination of global fit statistics. We then built three auxiliary models using the three-step Bolck-Croon-Hagenaars (BCH) method to test the relationship of latent class membership to several hypothesized social and health variables in this same three month time period. Results Fit statistics, minimum classification probabilities, and ease of interpretation indicated a three-class solution for level of material difficulty. We termed these classes "High Difficulty" (n = 82), "Mixed Difficulty" (n = 215), and "Low Difficulty" (n = 98). Average classification probabilities indicated good class separability. "High Difficulty" participants had high probabilities of usually having difficulty accessing all five resources. "Mixed Difficulty" participants indicated a range of difficulty accessing all resources, with restrooms and bathing facilities being the most difficult. "Low Difficulty" participants were defined by high probabilities of never having access difficulty. In auxiliary analyses, there were significant (p < 0.05) differences in experiences of displacement, opioid withdrawal symptoms, nonfatal overdose, and violent victimization between classes. Conclusions This LCA indicates that among PWUD experiencing homelessness there exist distinct differences in resource access and material hardship, and that these differences are linked with political, social, substance use, and other health outcomes. We add to the literature on the relationship between poverty and health among PWUD. Policies which increase difficulty accessing necessary material resources may negatively impact health in this population.

Opioid use disorder (OUD) remains a major public health crisis in the United States, with significant challenges in treatment access, retention, and workforce capacity. OUD care teams, including addiction medicine physicians and peer recovery coaches (PRCs), support patients receiving medication for OUD (MOUD), yet face heavy workloads and burnout. Artificial intelligence (AI) innovations, particularly large language model (LLM)–powered chatbots, may extend PRC support and provide patients with on-demand recovery support between clinic visits and PRC contacts. However, evidence on their development, feasibility, acceptability, and usability in addiction services is limited. To describe the development, feasibility, acceptability, and usability of an AI-powered health coaching chatbot (“Suzy”) designed to support patients in OUD recovery. Clinicians, researchers, and technology developers conducted a multiphase study. In the formative phase, we conducted focus groups and interviews with 12 health care professionals and 8 patients with substance use histories to specify chatbot functions and then developed a rule-based chatbot. In the pilot phase, we conducted usability testing on the rule-based chatbot with 8 patients reporting substance use, who completed standardized tasks, surveys, and qualitative interviews. Measures included the System Usability Scale (SUS), Net Promoter Score (NPS), and Single Ease of Use Question (SEQ). In the LLM development phase, we developed a LLM chatbot co-designed with PRCs, OUD recovery patients and other substance use experts. Chatbot functions included craving management, appointment reminders, and resource referrals. All usability testing tasks were completed, supporting feasibility. Quantitative and qualitative feedback indicated strong acceptability and usability with an average SUS score of 93 (benchmark 68), NPS of 63 (benchmark 35), and mean SEQ score of 6.5/7. Patients valued Suzy’s approachable, nonjudgmental language, and features that promoted accountability and self-reflection and 24/7 availability, while emphasizing that chatbots should supplement but not replace human support. The LLM-enhanced chatbot development emphasized safety, accuracy, safety escalation protocols to mitigate risks of inappropriate chatbot responses, human-in-the-loop features, and expanded conversational flexibility and personal tailoring. A rule-based chatbot, designed to support OUD care, demonstrated strong feasibility, usability, and acceptability. LLM chatbot development required more robust safety and emergency reporting features while having more patient-responsive conversational functions. By providing on-demand coaching, referrals, and reminders, Suzy may extend the reach of care teams, alleviate provider burden, and enhance patient engagement. Additional work is needed to understand how to best integrate Suzy into the patient’s recovery journey to ensure human support remains accessible and prioritized. Next steps include evaluating the use of Suzy after LLM integration in real-world settings and its efficacy in reducing substance use.