An open source knowledge graph ecosystem for the life sciences.

Translational research requires data at multiple scales of biological organization. Advancements in sequencing and multi-omics technologies have increased the availability of these data, but researchers face significant integration challenges. Knowledge graphs (KGs) are used to model complex phenomena, and methods exist to construct them automatically. However, tackling complex biomedical integration problems requires flexibility in the way knowledge is modeled. Moreover, existing KG construction methods provide robust tooling at the cost of fixed or limited choices among knowledge representation models. PheKnowLator (Phenotype Knowledge Translator) is a semantic ecosystem for automating the FAIR (Findable, Accessible, Interoperable, and Reusable) construction of ontologically grounded KGs with fully customizable knowledge representation. The ecosystem includes KG construction resources (e.g., data preparation APIs), analysis tools (e.g., SPARQL endpoint resources and abstraction algorithms), and benchmarks (e.g., prebuilt KGs). We evaluated the ecosystem by systematically comparing it to existing open-source KG construction methods and by analyzing its computational performance when used to construct 12 different large-scale KGs. With flexible knowledge representation, PheKnowLator enables fully customizable KGs without compromising performance or usability.

Targeting Wnt signaling for improved glioma immunotherapy.

Introduction: Despite aggressive standard-of-care therapy, including surgery, radiation, and chemotherapy, glioblastoma recurrence is almost inevitable and uniformly lethal. Activation of glioma-intrinsic Wnt/ß-catenin signaling is associated with a poor prognosis and the proliferation of glioma stem-like cells, leading to malignant transformation and tumor progression. Impressive results in a subset of cancers have been obtained using immunotherapies including anti-CTLA4, anti-PD-1, and anti-PD-L1 or chimeric antigen receptor (CAR) T cell therapies. However, the heterogeneity of tumors, low mutational burden, single antigen targeting, and associated antigen escape contribute to non-responsiveness and potential tumor recurrence despite these therapeutic efforts. In the current study, we determined the effects of the small molecule, highly specific Wnt/CBP (CREB Binding Protein)/ß-catenin antagonist ICG-001, on glioma tumor cells and the tumor microenvironment (TME)-including its effect on immune cell infiltration, blood vessel decompression, and metabolic changes. Methods: Using multiple glioma patient-derived xenografts cell lines and murine tumors (GL261, K-Luc), we demonstrated in vitro cytostatic effects and a switch from proliferation to differentiation after treatment with ICG-001. Results: In these glioma cell lines, we further demonstrated that ICG-001 downregulated the CBP/ß-catenin target gene Survivin/BIRC5-a hallmark of Wnt/CBP/ß-catenin inhibition. We found that in a syngeneic mouse model of glioma (K-luc), ICG-001 treatment enhanced tumor infiltration by CD3+ and CD8+ cells with increased expression of the vascular endothelial marker CD31 (PECAM-1). We also observed differential gene expression and induced immune cell infiltration in tumors pretreated with ICG-001 and then treated with CAR T cells as compared with single treatment groups or when ICG-001 treatment was administered after CAR T cell therapy. Discussion: We conclude that specific Wnt/CBP/ß-catenin antagonism results in pleotropic changes in the glioma TME, including glioma stem cell differentiation, modulation of the stroma, and immune cell activation and recruitment, thereby suggesting a possible role for enhancing immunotherapy in glioma patients.

Androgen deprivation induces double-null prostate cancer via aberrant nuclear export and ribosomal biogenesis through HGF and Wnt activation.

Androgen deprivation therapy (ADT) targeting androgen/androgen receptor (AR)- signaling pathways is the main therapy for advanced prostate cancer (PCa). However, ADT eventually fails in most patients who consequently develop castration-resistant prostate cancer (CRPC). While more potent AR antagonists and blockers for androgen synthesis were developed to improve clinical outcomes, they also show to induce more diverse CRPC phenotypes. Specifically, the AR- and neuroendocrine-null PCa, DNPC, occurs in abiraterone and enzalutamide-treated patients. Here, we uncover that current ADT induces aberrant HGF/MET signaling activation that further elevates Wnt/ß-catenin signaling in human DNPC samples. Co-activation of HGF/MET and Wnt/ß-catenin axes in mouse prostates induces DNPC-like lesions. Single-cell RNA sequencing analyses identify increased expression and activity of XPO1 and ribosomal proteins in mouse DNPC-like cells. Elevated expression of XPO1 and ribosomal proteins is also identified in clinical DNPC specimens. Inhibition of XPO1 and ribosomal pathways represses DNPC growth in both in vivo and ex vivo conditions, evidencing future therapeutic targets.

Building a vertically integrated genomic learning health system: The biobank at the Colorado Center for Personalized Medicine.

Precision medicine initiatives across the globe have led to a revolution of repositories linking large-scale genomic data with electronic health records, enabling genomic analyses across the entire phenome. Many of these initiatives focus solely on research insights, leading to limited direct benefit to patients. We describe the biobank at the Colorado Center for Personalized Medicine (CCPM Biobank) that was jointly developed by the University of Colorado Anschutz Medical Campus and UCHealth to serve as a unique, dual-purpose research and clinical resource accelerating personalized medicine. This living resource currently has more than 200,000 participants with ongoing recruitment. We highlight the clinical, laboratory, regulatory, and HIPAA-compliant informatics infrastructure along with our stakeholder engagement, consent, recontact, and participant engagement strategies. We characterize aspects of genetic and geographic diversity unique to the Rocky Mountain region, the primary catchment area for CCPM Biobank participants. We leverage linked health and demographic information of the CCPM Biobank participant population to demonstrate the utility of the CCPM Biobank to replicate complex trait associations in the first 33,674 genotyped individuals across multiple disease domains. Finally, we describe our current efforts toward return of clinical genetic test results, including high-impact pathogenic variants and pharmacogenetic information, and our broader goals as the CCPM Biobank continues to grow. Bringing clinical and research interests together fosters unique clinical and translational questions that can be addressed from the large EHR-linked CCPM Biobank resource within a HIPAA- and CLIA-certified environment.

Inhibition of Wnt/ß-catenin signaling reduces renal fibrosis in murine glycogen storage disease type Ia.

Glycogen storage disease type Ia (GSD-Ia) is caused by a deficiency in the enzyme glucose-6-phosphatase-α (G6Pase-α or G6PC) that is expressed primarily in the gluconeogenic organs, namely liver, kidney cortex, and intestine. Renal G6Pase-α deficiency in GSD-Ia is characterized by impaired gluconeogenesis, nephromegaly due to elevated glycogen accumulation, and nephropathy caused, in part, by renal fibrosis, mediated by activation of the renin-angiotensin system (RAS). The Wnt/ß-catenin signaling regulates the expression of a variety of downstream mediators implicated in renal fibrosis, including multiple genes in the RAS. Sustained activation of Wnt/ß-catenin signaling is associated with the development and progression of renal fibrotic lesions that can lead to chronic kidney disease. In this study, we examined the molecular mechanism underlying GSD-Ia nephropathy. Damage to the kidney proximal tubules is known to trigger acute kidney injury (AKI) that can, in turn, activate Wnt/ß-catenin signaling. We show that GSD-Ia mice have AKI that leads to activation of the Wnt/ß-catenin/RAS axis. Renal fibrosis was demonstrated by increased renal levels of Snail1, α-smooth muscle actin (α-SMA), and extracellular matrix proteins, including collagen-Iα1 and collagen-IV. Treating GSD-Ia mice with a CBP/ß-catenin inhibitor, ICG-001, significantly decreased nuclear translocated active ß-catenin and reduced renal levels of renin, Snail1, α-SMA, and collagen-IV. The results suggest that inhibition of Wnt/ß-catenin signaling may be a promising therapeutic strategy for GSD-Ia nephropathy.

MENDS-on-FHIR: Leveraging the OMOP common data model and FHIR standards for national chronic disease surveillance.

Objective: The Multi-State EHR-Based Network for Disease Surveillance (MENDS) is a population-based chronic disease surveillance distributed data network that uses institution-specific extraction-transformation-load (ETL) routines. MENDS-on-FHIR examined using Health Language Seven's Fast Healthcare Interoperability Resources (HL7® FHIR®) and US Core Implementation Guide (US Core IG) compliant resources derived from the Observational Medical Outcomes Partnership (OMOP) Common Data Model (CDM) to create a standards-based ETL pipeline. Materials and Methods: The input data source was a research data warehouse containing clinical and administrative data in OMOP CDM Version 5.3 format. OMOP-to-FHIR transformations, using a unique JavaScript Object Notation (JSON)-to-JSON transformation language called Whistle, created FHIR R4 V4.0.1/US Core IG V4.0.0 conformant resources that were stored in a local FHIR server. A REST-based Bulk FHIR $export request extracted FHIR resources to populate a local MENDS database. Results: Eleven OMOP tables were used to create 10 FHIR/US Core compliant resource types. A total of 1.13 trillion resources were extracted and inserted into the MENDS repository. A very low rate of non-compliant resources was observed. Discussion: OMOP-to-FHIR transformation results passed validation with less than a 1% non-compliance rate. These standards-compliant FHIR resources provided standardized data elements required by the MENDS surveillance use case. The Bulk FHIR application programming interface (API) enabled population-level data exchange using interoperable FHIR resources. The OMOP-to-FHIR transformation pipeline creates a FHIR interface for accessing OMOP data. Conclusion: MENDS-on-FHIR successfully replaced custom ETL with standards-based interoperable FHIR resources using Bulk FHIR. The OMOP-to-FHIR transformations provide an alternative mechanism for sharing OMOP data.

Reprogramming of 3D chromatin domains by antagonizing the ß-catenin/CBP interaction attenuates insulin signaling in pancreatic cancer.

The therapeutic potential of targeting the ß-catenin/CBP interaction has been demonstrated in a variety of preclinical tumor models with a small molecule inhibitor, ICG-001, characterized as a ß-catenin/CBP antagonist. Despite the high binding specificity of ICG-001 for the N-terminus of CBP, this ß-catenin/CBP antagonist exhibits pleiotropic effects. Our recent studies found global changes in three-dimensional (3D) chromatin architecture in response to disruption of the ß-catenin/CBP interaction in pancreatic cancer cells. However, an understanding of the functional crosstalk between antagonizing the ß-catenin/CBP interaction effect changes in 3D chromatin architecture and thereby gene expression and downstream effects remains to be elucidated. Here we perform Hi-C analyses on canonical and patient-derived pancreatic cancer cells before and after the treatment with ICG-001. In addition to global alteration of 3D chromatin domains, we unexpectedly identify insulin signaling genes enriched in the altered chromatin domains. We further demonstrate the chromatin loops associated with insulin signaling genes are significantly weakened after ICG-001 treatment. We finally elicit the deletion of a looping of IRS1, a key insulin signaling gene, significantly impede pancreatic cancer cell growth, indicating that looping-mediated insulin signaling might act as an oncogenic pathway to promote pancreatic cancer progression. Our work shows that targeting aberrant insulin chromatin looping in pancreatic cancer might provide a therapeutic benefit.

Sustained Effect of Clinical Decision Support for Heart Failure: A Natural Experiment Using Implementation Science.

OBJECTIVES: In a randomized controlled trial, we found that applying implementation science (IS) methods and best practices in clinical decision support (CDS) design to create a locally customized, "enhanced" CDS significantly improved evidence-based prescribing of ß blockers (BB) for heart failure compared with an unmodified commercially available CDS. At trial conclusion, the enhanced CDS was expanded to all sites. The purpose of this study was to evaluate the real-world sustained effect of the enhanced CDS compared with the commercial CDS. METHODS: In this natural experiment of 28 primary care clinics, we compared clinics exposed to the commercial CDS (preperiod) to clinics exposed to the enhanced CDS (both periods). The primary effectiveness outcome was the proportion of alerts resulting in a BB prescription. Secondary outcomes included patient reach and clinician adoption (dismissals). RESULTS: There were 367 alerts for 183 unique patients and 171 unique clinicians (pre: March 2019-August 2019; post: October 2019-March 2020). The enhanced CDS increased prescribing by 26.1% compared with the commercial (95% confidence interval [CI]: 17.0-35.1%), which is consistent with the 24% increase in the previous study. The odds of adopting the enhanced CDS was 81% compared with 29% with the commercial (odds ratio: 4.17, 95% CI: 1.96-8.85). The enhanced CDS adoption and effectiveness rates were 62 and 14% in the preperiod and 92 and 10% in the postperiod. CONCLUSION: Applying IS methods with CDS best practices was associated with improved and sustained clinician adoption and effectiveness compared with a commercially available CDS tool.

Vertebral osteomyelitis secondary to Streptococcus cristatus infection.

A 66-year-old male with a history of low back pain was found to have discitis and osteomyelitis. Biopsy and PCR testing revealed Streptococcus cristatus infection. This bacteria does not typically cause disease, and only a few cases in the literature have reported it to cause infection in the bones or joints. This case illustrates that vertebral osteomyelitis with a rare causative agent, S. cristatus, is possible and can be identified with PCR. Treatment typically requires long-term antibiotics tailored to the causative agent for a minimum of 6 weeks and can sometimes include surgical management.

Ontologizing health systems data at scale: making translational discovery a reality.

Common data models solve many challenges of standardizing electronic health record (EHR) data but are unable to semantically integrate all of the resources needed for deep phenotyping. Open Biological and Biomedical Ontology (OBO) Foundry ontologies provide computable representations of biological knowledge and enable the integration of heterogeneous data. However, mapping EHR data to OBO ontologies requires significant manual curation and domain expertise. We introduce OMOP2OBO, an algorithm for mapping Observational Medical Outcomes Partnership (OMOP) vocabularies to OBO ontologies. Using OMOP2OBO, we produced mappings for 92,367 conditions, 8611 drug ingredients, and 10,673 measurement results, which covered 68-99% of concepts used in clinical practice when examined across 24 hospitals. When used to phenotype rare disease patients, the mappings helped systematically identify undiagnosed patients who might benefit from genetic testing. By aligning OMOP vocabularies to OBO ontologies our algorithm presents new opportunities to advance EHR-based deep phenotyping.

Contextualising adverse events of special interest to characterise the baseline incidence rates in 24 million patients with COVID-19 across 26 databases: a multinational retrospective cohort study.

Background: Adverse events of special interest (AESIs) were pre-specified to be monitored for the COVID-19 vaccines. Some AESIs are not only associated with the vaccines, but with COVID-19. Our aim was to characterise the incidence rates of AESIs following SARS-CoV-2 infection in patients and compare these to historical rates in the general population. Methods: A multi-national cohort study with data from primary care, electronic health records, and insurance claims mapped to a common data model. This study's evidence was collected between Jan 1, 2017 and the conclusion of each database (which ranged from Jul 2020 to May 2022). The 16 pre-specified prevalent AESIs were: acute myocardial infarction, anaphylaxis, appendicitis, Bell's palsy, deep vein thrombosis, disseminated intravascular coagulation, encephalomyelitis, Guillain- Barré syndrome, haemorrhagic stroke, non-haemorrhagic stroke, immune thrombocytopenia, myocarditis/pericarditis, narcolepsy, pulmonary embolism, transverse myelitis, and thrombosis with thrombocytopenia. Age-sex standardised incidence rate ratios (SIR) were estimated to compare post-COVID-19 to pre-pandemic rates in each of the databases. Findings: Substantial heterogeneity by age was seen for AESI rates, with some clearly increasing with age but others following the opposite trend. Similarly, differences were also observed across databases for same health outcome and age-sex strata. All studied AESIs appeared consistently more common in the post-COVID-19 compared to the historical cohorts, with related meta-analytic SIRs ranging from 1.32 (1.05 to 1.66) for narcolepsy to 11.70 (10.10 to 13.70) for pulmonary embolism. Interpretation: Our findings suggest all AESIs are more common after COVID-19 than in the general population. Thromboembolic events were particularly common, and over 10-fold more so. More research is needed to contextualise post-COVID-19 complications in the longer term. Funding: None.

CDK5RAP2 is a Wnt target gene and promotes stemness and progression of oral squamous cell carcinoma.

In oral squamous cell carcinoma (OSCC), a highly aggressive and frequently lethal malignancy, the role and action mechanism of the microtubule regulatory protein CDK5RAP2 have not been fully understood. Here, we show that CDK5RAP2 is highly expressed in OSCC and its expression correlates with clinical stage and lymph node metastasis of the disease. The expression of CDK5RAP2 is regulated by the Wnt signaling pathway. Depletion of CDK5RAP2 inhibits the tumorigenesis and migration of OSCC cells and alters the OSCC cancer stem (-like) cell (CSC) signature. Notably, suppression of CDK5RAP2 expression disrupts spindle orientation during mitosis. Collectively, these results identify CDK5RAP2 as a potential CSC marker and reveal a mechanism that controls the CSC population in OSCC.

A Metabolic Smorgasbord Drives and Sustains CSC Heterogeneity.

A deeper understanding of the biology of therapy resistance is important for the development of optimal strategies to attain complete cancer cures [...].

Prevention and Management of Chronic Postsurgical Pain and Persistent Opioid Use Following Solid Organ Transplantation: Experiences From the Toronto General Hospital Transitional Pain Service.

BACKGROUND: With >700 transplant surgeries performed each year, Toronto General Hospital (TGH) is currently one of the largest adult transplant centers in North America. There is a lack of literature regarding both the identification and management of chronic postsurgical pain (CPSP) after organ transplantation. Since 2014, the TGH Transitional Pain Service (TPS) has helped manage patients who developed CPSP after solid organ transplantation (SOT), including heart, lung, liver, and renal transplants. METHODS: In this retrospective cohort study, we describe the association between opioid consumption, psychological characteristics of pain, and demographic characteristics of 140 SOT patients who participated in the multidisciplinary treatment at the TGH TPS, incorporating psychology and physiotherapy as key parts of our multimodal pain management regimen. RESULTS: Treatment by the multidisciplinary TPS team was associated with significant improvement in pain severity and a reduction in opioid consumption. CONCLUSIONS: Given the risk of CPSP after SOT, robust follow-up and management by a multidisciplinary team should be considered to prevent CPSP, help guide opioid weaning, and provide psychological support to these patients to improve their recovery trajectory and quality of life postoperatively.

ReviewR: a light-weight and extensible tool for manual review of clinical records.

Objectives: Manual record review is a crucial step for electronic health record (EHR)-based research, but it has poor workflows and is error prone. We sought to build a tool that provides a unified environment for data review and chart abstraction data entry. Materials and Methods: ReviewR is an open-source R Shiny application that can be deployed on a single machine or made available to multiple users. It supports multiple data models and database systems, and integrates with the REDCap API for storing abstraction results. Results: We describe 2 real-world uses and extensions of ReviewR. Since its release in April 2021 as a package on CRAN it has been downloaded 2204 times. Discussion and Conclusion: ReviewR provides an easily accessible review interface for clinical data warehouses. Its modular, extensible, and open source nature afford future expansion by other researchers.

A process to deduplicate individuals for regional chronic disease prevalence estimates using a distributed data network of electronic health records.

Introduction: Learning health systems can help estimate chronic disease prevalence through distributed data networks (DDNs). Concerns remain about bias introduced to DDN prevalence estimates when individuals seeking care across systems are counted multiple times. This paper describes a process to deduplicate individuals for DDN prevalence estimates. Methods: We operationalized a two-step deduplication process, leveraging health information exchange (HIE)-assigned network identifiers, within the Colorado Health Observation Regional Data Service (CHORDS) DDN. We generated prevalence estimates for type 1 and type 2 diabetes among pediatric patients (0-17 years) with at least one 2017 encounter in one of two geographically-proximate DDN partners. We assessed the extent of cross-system duplication and its effect on prevalence estimates. Results: We identified 218 437 unique pediatric patients seen across systems during 2017, including 7628 (3.5%) seen in both. We found no measurable difference in prevalence after deduplication. The number of cases we identified differed slightly by data reconciliation strategy. Concordance of linked patients' demographic attributes varied by attribute. Conclusions: We implemented an HIE-dependent, extensible process that deduplicates individuals for less biased prevalence estimates in a DDN. Our null pilot findings have limited generalizability. Overlap was small and likely insufficient to influence prevalence estimates. Other factors, including the number and size of partners, the matching algorithm, and the electronic phenotype may influence the degree of deduplication bias. Additional use cases may help improve understanding of duplication bias and reveal other principles and insights. This study informed how DDNs could support learning health systems' response to public health challenges and improve regional health.

The CBP/ß-Catenin Antagonist, ICG-001, Inhibits Tumor Metastasis via Blocking of the miR-134/ITGB1 Axis-Mediated Cell Adhesion in Nasopharyngeal Carcinoma.

Nasopharyngeal carcinoma (NPC) is an Epstein-Barr virus (EBV)-associated malignancy ranking as the 23rd most common cancer globally, while its incidence rate ranked the 9th in southeast Asia. Tumor metastasis is the dominant cause for treatment failure in NPC and metastatic NPC is yet incurable. The Wnt/ß-catenin signaling pathway plays an important role in many processes such as cell proliferation, differentiation, epithelial-mesenchymal transition (EMT), and self-renewal of stem cells and cancer stem cells (CSCs). Both the EMT process and CSCs are believed to play a critical role in cancer metastasis. We here investigated whether the specific CBP/ß-catenin Wnt antagonist, IGC-001, affects the metastasis of NPC cells. We found that ICG-001 treatment could reduce the adhesion capability of NPC cells to extracellular matrix and to capillary endothelial cells and reduce the tumor cell migration and invasion, events which are closely associated with distant metastasis. Through a screening of EMT and CSC-related microRNAs, it was found that miR-134 was consistently upregulated by ICG-001 treatment in NPC cells. Very few reports have mentioned the functional role of miR-134 in NPC, except that the expression was found to be downregulated in NPC. Transient transfection of miR-134 into NPC cells reduced their cell adhesion, migration, and invasion capability, but did not affect the growth of CSC-enriched tumor spheres. Subsequently, we found that the ICG-001-induced miR-134 expression resulting in downregulation of integrin ß1 (ITGB1). Such downregulation reduced cell adhesion and migration capability, as demonstrated by siRNA-mediated knockdown of ITGB1. Direct targeting of ITGB1 by miR-134 was confirmed by the 3'-UTR luciferase assay. Lastly, using an in vivo lung metastasis assay, we showed that ICG-001 transient overexpression of miR-134 or stable overexpression of miR-134 could significantly reduce the lung metastasis of NPC cells. Taken together, we present here evidence that modulation of Wnt/ß-catenin signaling pathway could inhibit the metastasis of NPC through the miR-134/ITGB1 axis.

An Exploratory Study of the Social Contract of Research and Innovation in Africa.

This exploratory study adds to the under-developed literature on a Research Topic that laden with epistemological, philosophical, and ideological overtones, and that begs many questions. The literature on political economy generally, and that for Africa, enjoys full disciplinary status. In contrast the political economy of research and innovation remains an emerging interdisciplinary field that examines the overlap between innovation studies and political economy. The pursuit of "science and technology" was expected to play its part in the imperialist and colonial agendas, and in the post-colonial project, when science and technology policy was a strong element in advocacy for Africa's post-independence development. What have the policies achieved, and what explains the shortfalls? What indeed is the relationship between industrial policy and research and innovation policy? What is the social contract with research and innovation? The study commences with a general overview of the social contract for science before turning to Africa's post-independence modernizing agenda, and the roles ascribed to industrial policy and research and innovation policy. An eclectic methodology drawing on Cloutier (2021) is deployed to characterize and measure the social contract between research and innovation. The methodology adapts Cloutier (2021) to the functionality of national innovation systems. The responsiveness of STI policy is further probed using Martin, 2015 categorization of innovation policy informed by Theory of Change. Where possible reference is made to conventional STI indicators. Research and innovation policy is then assessed at continental and national levels, with attention given to the extent of linkages in national innovation systems. Further to tease out the various forms of social contract, five country-level STI policies are analyzed using the Martin categorization and Theory of Change methodology. It will be argued that a binding, social contract for inclusive research and innovation policy is largely absent, so that the prospects for attaining the SDGs remain elusive. Post Glasgow COP-26, donor pressure might be re-oriented to promote engagement with the Sustainable Development Goals, though the upholding research sovereignty could mitigate against this. Africa might rightly chide against such pressure, given her experience of what has been labeled as "vaccine apartheid."

Identifying who has long COVID in the USA: a machine learning approach using N3C data.

BACKGROUND: Post-acute sequelae of SARS-CoV-2 infection, known as long COVID, have severely affected recovery from the COVID-19 pandemic for patients and society alike. Long COVID is characterised by evolving, heterogeneous symptoms, making it challenging to derive an unambiguous definition. Studies of electronic health records are a crucial element of the US National Institutes of Health's RECOVER Initiative, which is addressing the urgent need to understand long COVID, identify treatments, and accurately identify who has it-the latter is the aim of this study. METHODS: Using the National COVID Cohort Collaborative's (N3C) electronic health record repository, we developed XGBoost machine learning models to identify potential patients with long COVID. We defined our base population (n=1 793 604) as any non-deceased adult patient (age ≥18 years) with either an International Classification of Diseases-10-Clinical Modification COVID-19 diagnosis code (U07.1) from an inpatient or emergency visit, or a positive SARS-CoV-2 PCR or antigen test, and for whom at least 90 days have passed since COVID-19 index date. We examined demographics, health-care utilisation, diagnoses, and medications for 97 995 adults with COVID-19. We used data on these features and 597 patients from a long COVID clinic to train three machine learning models to identify potential long COVID among all patients with COVID-19, patients hospitalised with COVID-19, and patients who had COVID-19 but were not hospitalised. Feature importance was determined via Shapley values. We further validated the models on data from a fourth site. FINDINGS: Our models identified, with high accuracy, patients who potentially have long COVID, achieving areas under the receiver operator characteristic curve of 0·92 (all patients), 0·90 (hospitalised), and 0·85 (non-hospitalised). Important features, as defined by Shapley values, include rate of health-care utilisation, patient age, dyspnoea, and other diagnosis and medication information available within the electronic health record. INTERPRETATION: Patients identified by our models as potentially having long COVID can be interpreted as patients warranting care at a specialty clinic for long COVID, which is an essential proxy for long COVID diagnosis as its definition continues to evolve. We also achieve the urgent goal of identifying potential long COVID in patients for clinical trials. As more data sources are identified, our models can be retrained and tuned based on the needs of individual studies. FUNDING: US National Institutes of Health and National Center for Advancing Translational Sciences through the RECOVER Initiative.