About

Emily Hartwell, PhD, is an Assistant Professor of Psychiatry at the Hospital of the University of Pennsylvania and an Associate Member of the Institute for Translational Medicine and Therapeutics Research. She is also a psychologist at the Crescenz Veterans Affairs Medical Center affiliated with the University of Pennsylvania. Her educational background includes a BA in Political Science from North Carolina State University, an MA in Psychology from the University of California, Los Angeles, and a PhD in Clinical Psychology from UCLA. Her research focuses on substance use disorders, with a particular emphasis on alcohol use disorder. She has contributed to understanding the genetic and pharmacogenetic factors influencing substance use and treatment responses. Her work includes systematic reviews, meta-analyses, and phenome-wide association analyses related to substance use, genetics, and treatment outcomes. Hartwell has been an active member of professional societies such as the Society on Alcohol and the College on Problems of Drug Dependence since 2010 and 2018, respectively.

Research topics

• Medicine
• Psychiatry
• Psychology
• Clinical psychology
• Internal medicine

Selected publications

• Predicting treatment-seeking status for alcohol use disorder using polygenic scores and machine learning in a deeply-phenotyped sample

  Drug and Alcohol Dependence · 2025-07-16 · 1 citations

  articleOpen accessSenior authorCorresponding
  PublisherOA PDFDOI

• Chapter 3. Alcohol Use Disorder

  American Psychiatric Association Publishing eBooks · 2025-11-01

  book-chapterSenior author
  PublisherDOI

• Utility of Candidate Genes From an Algorithm Designed to Predict Genetic Risk for Opioid Use Disorder

  JAMA Network Open · 2025-01-09 · 5 citations

  articleOpen access

  Importance: Recently, the US Food and Drug Administration gave premarketing approval to an algorithm based on its purported ability to identify individuals at genetic risk for opioid use disorder (OUD). However, the clinical utility of the candidate genetic variants included in the algorithm has not been independently demonstrated. Objective: To assess the utility of 15 genetic variants from an algorithm intended to predict OUD risk. Design, Setting, and Participants: This case-control study examined the association of 15 candidate genetic variants with risk of OUD using electronic health record data from December 20, 1992, to September 30, 2022. Electronic health record data, including pharmacy records, were accrued from participants in the Million Veteran Program across the US with opioid exposure (n = 452 664). Cases with OUD were identified using International Classification of Diseases, Ninth Revision, or International Classification of Diseases, Tenth Revision, diagnostic codes, and controls were individuals with no OUD diagnosis. Exposures: Number of risk alleles present across 15 candidate genetic variants. Main Outcome and Measures: Performance of 15 genetic variants for identifying OUD risk assessed via logistic regression and machine learning models. Results: A total of 452 664 individuals with opioid exposure (including 33 669 with OUD) had a mean (SD) age of 61.15 (13.37) years, and 90.46% were male; the sample was ancestrally diverse (with individuals of genetically inferred European, African, and admixed American ancestries). Using Nagelkerke R2, collectively, the 15 candidate genes accounted for 0.40% of variation in OUD risk. In comparison, age and sex alone accounted for 3.27% of the variation. The ensemble machine learning. The ensemble machine learning model using the 15 variants as predictive factors correctly classified 52.83% (95% CI, 52.07%-53.59%) of individuals in an independent testing sample. Conclusions and Relevance: Results of this study suggest that the candidate genetic variants included in the approved algorithm do not meet reasonable standards of efficacy in identifying OUD risk. Given the algorithm's limited predictive accuracy, its use in clinical care would lead to high rates of both false-positive and false-negative findings. More clinically useful models are needed to identify individuals at risk of developing OUD.

  PublisherOA PDFDOI

• 26. X, Y, AND WHY: UNPACKING SEX DIFFERENCES IN ALCOHOL USE AND DEPRESSION

  European Neuropsychopharmacology · 2025-10-01

  article1st authorCorresponding
  PublisherDOI

• Multi-ancestry genome-wide association meta-analysis of buprenorphine treatment response

  Neuropsychopharmacology · 2025-05-06 · 1 citations

  reviewOpen access

  Abstract Although the mu-opioid partial agonist buprenorphine is increasingly being prescribed to treat opioid use disorder, patients’ responses to the drug vary and few clinical and no genetic predictors of treatment response have been identified. We conducted a genome-wide association study (GWAS) meta-analysis of buprenorphine treatment response (defined using urine drug screen results) in 4394 Veterans with opioid use disorder from the VA Million Veteran Program (751 of African-like ancestry [AFR] and 3643 of European-like ancestry [EUR]) and 296 participants from a clinical trial of extended-release buprenorphine (n AFR = 104, n EUR = 192). We conducted within-ancestry GWAS in both cohorts, followed by cross-ancestry, fixed-effects GWAS meta-analyses within and across cohorts. We also examined associations between demographic and clinical characteristics and buprenorphine treatment response. The cross-ancestry meta-analysis of both cohorts identified one genome-wide significant locus (rs149319538 ) that maps to SLC39A10 , a gene that encodes a zinc transporter. Phenome-wide association analyses of the lead variant implicated connectivity of the uncinate fasciculus, a limbic white matter fiber tract. Of the clinical characteristics, only the presence of chronic pain and a lower maximum buprenorphine dosage were related to higher odds of treatment response in adjusted models. We report here the first genome-wide significant variant associated with buprenorphine treatment response. Larger samples are needed to replicate these findings and identify additional clinical and genetic factors that predict buprenorphine treatment efficacy to enable the use of a precision approach to OUD treatment.

  PublisherOA PDFDOI

• Transdiagnostic and Disorder-Level Genome-Wide Association Studies Enhance Precision of Substance Use and Psychiatric Genetic Risk Profiles in African and European Ancestries

  Biological Psychiatry · 2025-05-08 · 6 citations

  articleOpen access

  BACKGROUND: Substance use disorders (SUDs) and psychiatric disorders frequently co-occur, and their etiology likely reflects both transdiagnostic (i.e., common/shared) and disorder-level (i.e., independent/nonshared) genetic influences. Understanding the genetic influences that are shared and those that operate independently of the shared risk could enhance precision in diagnosis, prevention, and treatment, but this remains underexplored, particularly in non-European ancestry groups. METHODS: We applied genomic structural equation modeling to examine the common and independent genetic architecture among SUDs and psychotic, mood, and anxiety disorders using summary statistics from genome-wide association studies (GWASs) conducted in European ancestry (EUR) and African ancestry (AFR) individuals. To characterize the biological and phenotypic associations, we used FUMA, conducted genetic correlations, and performed phenome-wide association studies (PheWASs). RESULTS: In EUR individuals, transdiagnostic genetic factors represented SUDs, psychotic disorders, and mood/anxiety disorders, with a GWAS identifying 2 novel lead single nucleotide polymorphisms (SNPs) for the mood factor. In AFR individuals, genetic factors represented SUDs and psychiatric disorders, and a GWAS identified 1 novel lead SNP for the SUD factor. In EUR individuals, second-order factor models showed phenotypic and genotypic associations with a broad range of physical and mental health traits. Finally, genetic correlations and PheWASs highlighted how common and independent genetic factors for SUDs and psychotic disorders were differentially associated with psychiatric, sociodemographic, and medical phenotypes. CONCLUSIONS: Combining transdiagnostic and disorder-level genetic approaches can improve our understanding of co-occurring conditions and increase the specificity of genetic discovery, which is critical for identifying more effective prevention and treatment strategies to reduce the burden of these disorders.

  PublisherDOI

• Polygenic Scores for Psychiatric Traits Are Associated With Substance Use Phenotypes in a Deeply Phenotyped Sample

  Drug and Alcohol Dependence · 2025-02-01

  article1st authorCorresponding
  PublisherDOI

• Polygenic Risk for Substance Use and Its Associations with Comorbities in the Penn Medicine BioBank

  Scholarly Commons (University of Pennsylvania) · 2025-09-15

  otherOpen accessSenior author

  The Penn Medicine BioBank (PMBB) contains the genetic information of 57,170 individuals linked to their electronic health records (EHRs) which store information on a patient’s diagnoses over time including substance abuse disorders (SUDs). Given that SUDs are highly heritable and polygenic, polygenic risk scores (PRS)—weighted sums of an individual’s genetic variants—are a method for estimating one’s genetic liability. We generated PRS for tobacco use disorder (TUD), alcohol use disorder (AUD), opioid use disorder (OUD), and cannabis use disorder (CanUD) using summary statistics from publicly available genome-wide association studies (GWAS). PRS was then calculated using PRS-CS and applied to individuals in the PMBB using PLINK. We evaluated primary associations between each PRS and its corresponding phenotype and we defined phenotype cases by 1+ inpatient International Classification of Disease (ICD) code or 2+ outpatient ICD code. We then conducted a phenome-wide association study (PheWAS) to discover the relationship between PRS and other clinical phenotypes. Primary phenotype associations were observed for PRSTUD (EUR and AFR), EUR PRSAUD, and EUR PRSCanUD (p < 0.05). PheWAS results revealed that EUR PRSAUD had significant positive associations with 15 phenotypes including TUD (p=2.7x10-41) and anxiety disorders (p=7.3x10-12) . EUR PRSCanUD, EUR PRSTUD, and AFR PRSTUD also had significant positive associations with TUD (p=5.9x10-38, p=8.0x10-47, and p=3.6x10-5, respectively). EUR PRSCanUD also had 17 more positive associations with different phenotypes including anxiety disorders (p=1.9x10-18), mood disorders (p=1.2x10-14), and viral hepatitis C (p=7.9x10-12). EUR PRSTUD also had positive associations with 31 more phenotypes including chronic airway obstruction (p=9.0x10-25), alongside 26 negative associations, most notably being benign neoplasm of skin (p=3.7x10-24). These findings support the highly polygenic nature of SUDs and demonstrate shared genetic liability with other psychiatric and clinical phenotypes.

  Publisher

• Demographic and Genetic Predictors of ADHD Diagnoses and Medications in Electronic Health Records

  Research Square · 2025-10-30

  preprintOpen accessSenior author
  PublisherOA PDFDOI

• Predicting Treatment-seeking Status for Alcohol Use Disorder Using Polygenic Scores and Machine Learning in a Deeply-Phenotyped Sample

  medRxiv · 2024-11-23

  preprintOpen accessSenior authorCorresponding

  Abstract Background Few individuals with alcohol use disorder (AUD) receive treatment. Previous studies have shown drinking behavior, psychological problems, and substance dependence to predict treatment seeking. However, to date, no studies have incorporated polygenic scores (PGS), a measure of genetic risk for AUD. Methods Using the Yale-Penn sample, we identified 9,103 individuals diagnosed with DSM-IV AUD and indicated treatment-seeking status. We implemented a random forest (RF) model to predict treatment-seeking based on 91 clinically relevant phenotypes. We calculated AUD PGS for those with genetic data (African ancestry [AFR] n=3,192, European ancestry [EUR] n=3,553) and generated RF models for each ancestry group, first without and then with PGS. Lastly, we developed models stratified by age (&lt; and ≥40 years old). Results 66.6% reported treatment seeking (M age =40.0, 62.4% male). Across models, top predictors included years of alcohol use and related psychological problems, psychiatric diagnoses, and heart disease. In the models without PGS, we found 79.8% accuracy and 0.85 AUC for EUR and 75% and 0.78 for AFR; the addition of PGS did not substantially change these metrics. PGS was the 10 th most important predictor for EUR and 23 rd for AFR. In the age-stratified analysis, PGS ranked 8 th for &lt;40 and 48 th for ≥40 in EUR ancestry, and it ranked 14 th for &lt;40 and 24 th for ≥40 in the AFR sample. Conclusion Alcohol use, psychiatric issues, and comorbid medical disorders were predictors of treatment seeking. Incorporating PGS did not substantially alter performance, but was a more important predictor in younger individuals with AUD. Highlights While alcohol use problems are common, few individuals seek treatment We used machine learning in a deeply-phenotyped sample to predict treatment-seeking We, for the first time, incorporated polygenic risk for alcohol use as a predictor Alcohol use variables, psychiatric issues, and medical problems were key predictors

  PublisherOA PDFDOI

Frequent coauthors

• Henry R. Kranzler

  Washington University in St. Louis

  174 shared

• Joel Gelernter

  137 shared

• Rachel L. Kember

  132 shared

• Amy C. Justice

  85 shared

• Sylvanus Toikumo

  82 shared

• Heng Xu

  Huaqiao University

  79 shared

• Rachel Vickers‐Smith

  University of Kentucky

  74 shared

• Divya Saini

  University of Pennsylvania

  71 shared

Education

• PhD, Psychology

  University of California Los Angeles

  2018

Awards & honors

• Member of the Society on Alcohol since 2018
• Member of the College on Problems of Drug Dependence since 2…