安舒茨医学中心

Large Language Models (LLMs) offer a promising avenue for scientific discovery, yet their application to symbolic regression is often constrained by inefficient search strategies and coarse feedback signals. Current methods typically guide LLMs using scalar metrics (e.g., global Mean Squared Error), which fail to identify which components of a proposed equation are driving performance or causing error. We introduce Influence-Guided Symbolic Regression (IGSR), a method that frames equation discovery as an iterative two-step process combining diverse term generation with rigorous selection: an LLM generates candidate basis functions ψ_j(𝐱) for a linear model, which are then evaluated using granular influence scores Δ_j. These scores quantify each term's marginal contribution to generalization accuracy, enabling an influence-guided pruning process that systematically refines the model structure. Integrating this mechanism into a Monte Carlo Tree Search (MCTS) enables navigating the combinatorial search space while balancing exploration of novel functional forms with exploitation of high-influence components. We demonstrate IGSR's effectiveness on a diverse suite of benchmarks, including LLM-SRBench, pharmacological PKPD models, an epidemiological simulation, and real-world genomic data. Notably, we validate the framework's capacity for genuine discovery in a case study using a high-dimensional biological dataset, in which IGSR identified a novel relationship between DNA methylation and RNA Polymerase II pausing; a hypothesis that was subsequently supported via wet-lab experimentation.

Objectives Advance care planning (ACP) is essential in supporting family caregivers of nursing home residents with dementia, but nursing home (NH) staff often lack training to engage in proactive ACP discussions. An embedded pragmatic clinical trial was conducted to test a structured ACP training for NH staff called the ACP Specialist Program. This study explores family caregivers’ experiences related to discussions with the ACP Specialist, as well as needs and challenges in making ACP decisions for NH residents living with dementia. Design Qualitative interviews. Setting and Participants 28 family caregivers of NH residents with dementia who had engaged in an ACP discussion with the ACP Specialist in the prior 3-month period. Methods Interviews were conducted, transcribed, and then coded by the research team using qualitative descriptive methods. Results Thirteen original codes were distilled into 8 codes reflecting 2 broad themes. The first theme focuses on family caregiver experiences as the surrogate decision maker and includes the caregiver's understanding of the surrogate decision-maker role; comfort with the role; challenges related to the role; and involvement of other family members. The second theme focuses on caregiver perspectives of ACP Specialists and the ACP discussion. It includes descriptions of the types of decisions in ACP conversations; the experience of support from the ACP Specialist; the role of the ACP Specialist as part of the NH team; and the role of other NH staff and clinicians in ACP. Conclusion and Implications Family caregiver experiences were positive overall because they believed that the resident's situation, preferences, and needs were known and acted on by the staff. Their reports of needs and challenges reinforce the importance of ACP training programs like the ACP Specialist that promote routine conversations in the NH to support family decision makers for persons living with dementia.

Osteoporosis and chronic kidney disease (CKD)–metabolic bone disease (MBD) (CKD-MBD) are increasingly recognized as overlapping conditions, particularly in the aging population. Declining renal function and skeletal fragility often coexist, because CKD-MBD may develop in a skeleton already compromised by preexisting osteoporosis. Adynamic bone, often resulting from excessive suppression of parathyroid hormone (PTH) and now a common form of renal osteodystrophy (ROD), may histologically resemble low-turnover osteoporosis; distinguishing between the 2 under light microscopy remains difficult, and reliable differentiation often depends on clinical context. Nevertheless, nephrologists and nonnephrologist bone specialists frequently work in parallel rather than in collaboration.This separation has contributed to persistent diagnostic gaps and fragmented management, especially in patients with advanced CKD. Advances in imaging, biochemical markers, and bone histomorphometry have improved insight into disease mechanisms; however, limitations in current diagnostic approaches remain. Osteoporosis therapies are frequently underused in CKD, despite growing evidence supporting efficacy and safety across a broader range of kidney function than previously assumed. Despite efforts to refine the definition of osteoporosis beyond bone mineral density (BMD) alone, clinical misclassification continues.Beyond skeletal health, vascular calcification (VC)—driven by disordered calcium–phosphate homeostasis—remains insufficiently prioritized in clinical decision-making, despite its strong association with cardiovascular morbidity and mortality in CKD. Emerging concepts, such as intermittent PTH administration, an established treatment in osteoporosis, illustrate the potential for interventions that may restore mineral balance and improve skeletal integrity in selected CKD populations. Whether such strategies can also favorably influence cardiovascular risk remains uncertain and warrant investigation. This integrated framework may improve interdisciplinary care.

Diabetes Technology & TherapeuticsVol. 26, No. S1 Original ArticlesFree AccessVirtual Clinics for Diabetes CareSatish K. Garg, Amanda H. Rewers, and Gurleen KaurSatish K. GargBarbara Davis Center for Diabetes, University of Colorado, Aurora, CO, USA.Search for more papers by this author, Amanda H. RewersBarbara Davis Center for Diabetes, University of Colorado, Aurora, CO, USA.Search for more papers by this author, and Gurleen KaurBarbara Davis Center for Diabetes, University of Colorado, Aurora, CO, USA.Search for more papers by this authorPublished Online:1 Mar 2024https://doi.org/10.1089/dia.2024.2501AboutSectionsPDF/EPUB Permissions & CitationsPermissionsDownload CitationsTrack CitationsAdd to favorites Back To Publication ShareShare onFacebookXLinked InRedditEmail IntroductionThe COVID‐19 pandemic did have a silver lining: the use of telehealth or virtual clinics across the health‐care sector, especially for people with diabetes. Before the COVID‐19 pandemic, most of the health care for people with diabetes was provided by in‐person visits. It is known that in‐person visits are inconvenient for patients, including the travel time and half to a full day off from work. Virtual visits could be cost‐effective for patients as well as the providers, assuming the reimbursement for virtual care is similar to office visits. In the United States, the insurance companies are threatening to consider a different compensation model for telehealth visits because they do not involve a physical examination by the provider or the use of facilities. In addition, in the United States providers cannot provide virtual care without being licensed in the state the patient is in at the time of the telehealth visit. These may not be relevant factors in the European Union and the United Kingdom.It is well recognized that people with long‐standing diabetes may not need to see a provider in person four times a year, even though that is the guideline followed by the regulators and the insurance companies. One can imagine that patients who are stable with their care with a hemoglobin A1c (HbA1c) of < 7% and a time in range (TIR) of > 70% with no long‐term complications could easily get away with seeing their providers once or twice a year. However, in the United States, if you have Medicare or Medicaid and you use an insulin pump and a sensor (hybrid closed‐loop system [HCL]), you are required to see a provider at least four times a year, and more than half the time those visits are not needed. Above all, if many of these visits could be provided through virtual care, the extra time the providers have might be used to take care of the needs of those who are high‐risk patients with higher HbA1c values and associated long‐term microvascular and macrovascular complications.We learned during the COVID‐19 pandemic lockdowns that virtual care can be easily provided, even for new onset patients with type 1 diabetes (T1D) and especially with the availability of new technologies that allow remote access to the data. In addition, during the pandemic providers learned how to safely and effectively institute remote care (through Microsoft Teams or Zoom, etc.) using data from continuous glucose monitors (CGMs) and HCL systems. Now most of the companies have created modules to help people initiate many of these systems by themselves.Obesity also is clearly related to the increasing prevalence of type 2 diabetes (T2D), and it is discussed at length in the new medications article in this year's ATTD Yearbook.For our article on virtual care, we have chosen 14 abstracts out of the thousands that were published from July 1, 2022, to June 30, 2023, and divided them into four categories: (1) the role of telehealth in obesity, (2) remote management of diabetic retinopathy, (3) the role of telehealth during pregnancy, and (4) economic considerations in telehealth. Even though we have included five abstracts on managing diabetes during pregnancy through telehealth, we feel that it is important to highlight the need of virtual care during pregnancy in this year's ATTD Yearbook, in addition to what Helen Murphy and Jennifer Yamamoto have contributed in their article.Key Articles ReviewedA Nutritional Web‐based Approach in Obesity and Diabetes before and during the COVID‐19 LockdownFraticelli F, Di Nicola M, Vitacolonna EJ Telemed Telecare2023; 29:91–102Reversing Type 2 Diabetes in a Primary Care‐Anchored eHealth Lifestyle Coaching Programme in Denmark: A Randomised Controlled TrialChristensen JR, Laursen DH, Lauridsen JT, Hesseldal L, Jakobsen PR, Nielsen JB, Søndergaard J, Brandt CJNutrients2022; 14:3424Use of Video‐based Telehealth Services Using a Mobile App for Workers in Underserved Areas during the COVID‐19 Pandemic: A Prospective Observational StudyPark HS, Jeong S, Chung HY, Soh JY, Hyun YH, Bang SH, Kim HSInt J Med Inform2022; 166:104844Findings from a Statewide TeleRetina Diabetic Retinopathy Screening Program in ArkansasShirey M, Kwok A, Jenkins H, Uwaydat SInt J Telemed Appl2023; 2023:3233803Effectiveness of Telemedicine Diabetic Retinopathy Screening in the USA: A Protocol for Systematic Review and Meta‐analysisPadilla Conde T, Robinson L, Vora P, Ware SL, Stromberg A, Bastos de Carvalho ASyst Rev2023; 12:48Tele‐ophthalmology for Age‐related Macular Degeneration during the COVID‐19 Pandemic and BeyondMintz J, Labiste C, DiCaro MV, McElroy E, Alizadeh R, Xu KJ Telemed Telecare2022; 28:670–679Implementation of Deep Learning Artificial Intelligence in Vision‐threatening Disease Screenings for an Underserved Community during COVID‐19Zhu A, Tailor P, Verma R, Zhang I, Schott B, Ye C, Szirth B, Habiel M, Khouri ASJ Telemed Telecare 2023;1357633X231158832.Telehealth Adoption during COVID‐19: Lessons Learned from Obstetric Providers in the Rocky Mountain WestHolman C, Glover A, McKay K, Gerard CTelemed Rep2023; 4:1–9Telehealth Utilization in High‐risk Pregnancies during COVID‐19Rayford MA, Morris JM, Phinehas R, Schneider E, Lund A, Baxley S, Wan JY, Goedecke PJ, Levi‐D'Ancona RTelemed Rep2023; 4:61–66Patient‐reported Benefits and Limitations of Mobile Health Technologies for Diabetes in Pregnancy: A Scoping ReviewSushko K, Menezes HT, Wang QR, Nerenberg K, Fitzpatrick‐Lewis D, Sherifali DCan J Diabetes2023; 47:102–113Exploring the Acceptability and Experience of Receiving Diabetes and Pregnancy Care via Telehealth during the COVID‐19 Pandemic: A Qualitative StudyKozica‐Olenski SL, Soldatos G, Marlow L, Cooray SD, Boyle JABMC Pregnancy Childbirth2022; 22:932Pre‐pregnancy Health of Women with Pre‐existing Diabetes or Previous Gestational Diabetes: Analysis of Pregnancy Risk Factors and Behavioural Data from a Digital ToolFlynn AC, Robertson M, Kavanagh K, Murphy HR, Forde R, Stephenson J, Poston L, White SLDiabetes Med2023; 40:e15008The Role of Virtual Triage in Improving Clinician Experience and Satisfaction: A Narrative ReviewGellert GA, Rasławska‐Socha J, Marcjasz N, Price T, Heyduk A, Mlodawska A, Kuszczyński K, Jędruch A, Orzechowski PTelemed Rep2023; 4:180–191Payment and Coverage Parity for Virtual Care and In‐person Care: How Do We Get There?Khera N, Knoedler M, Meier SK, TerKonda S, Williams RD, Wittich CM, Coffey JD, Demaerschalk BMTelemed Rep2023; 4:100–108OBESITY AND TELEHEALTHA Nutritional Web‐based Approach in Obesity and Diabetes before and during the COVID‐19 LockdownFraticelli F1, Di Nicola M2, Vitacolonna E11Department of Medicine and Aging, School of Medicine and Health Sciences, “G. d'Annunzio” University of Chieti‐Pescara, Italy; 2Laboratory of Biostatistics, Department of Medical, Oral and Biotechnological Sciences, “G. d'Annunzio” University of Chieti‐Pescara, ItalyJ Telemed Telecare2023;29:91–102Type 2 diabetes (T2D) and obesity are closely related conditions, and novel technologies have provided promising tools for their management. The present study compared a web‐based nutritional intervention versus a traditional one, employed both before and during Italy's lockdown period during the COVID‐19 outbreak in overweight and obese individuals with T2D or impaired glucose regulation.MethodsThis study randomly allocated 36 individuals into two arms: a traditional arm that provided face‐to‐face individual and group‐based interventions and a web arm that deployed both the in‐presence traditional approach together with intervention provided through web technologies. The outcomes were the data resulting from the comparison between the participants' anthropometric and clinical parameters as well as their PREDIMED (Prevención con Dieta Mediterránea) scores at baseline with those at 3 months (T3), 6 months (T6), and at lockdown.ResultsParticipants in the web arm showed a progressive reduction in weight and body mass index (BMI) from baseline to T6, then a minimal increase of both parameters during the lockdown. An improvement in these parameters compared with baseline was observed in the control participants during the lockdown. In both arms, compared with baseline the PREDIMED score improved at T6. Significant variations were observed considering weight (P < 0.001), BMI (P = 0.001), and PREDIMED scores (P = 0.023) over time.ConclusionsA short‐term nutritional web‐based intervention was effective and feasible in patients affected by T2D or impaired glucose regulation before and during the COVID‐19 pandemic.Reversing Type 2 Diabetes in a Primary Care‐anchored eHealth Lifestyle Coaching Programme in Denmark: A Randomised Controlled TrialChristensen JR1,2,3, Laursen DH4, Lauridsen JT5, Hesseldal L1, Jakobsen PR1, Nielsen JB1, Søndergaard J1, Brandt CJ1,61Research Unit of General Practice, Department of Public Health, The Faculty of Health Sciences, University of Southern Denmark, Odense, Denmark; 2User Perspectives and Community‐Based Interventions, Department of Public Health, The Faculty of Health Sciences, University of Southern Denmark, Odense, Denmark; 3The MOVE Unit, Research Unit of General Practice, The Faculty of Health, Aarhus University, Aarhus, Denmark; 4Department of Public Health, Faculty of Health and Medical Science, University of Copenhagen, Copenhagen, Denmark; 5Department of Economics, The Faculty of Business and Social Sciences, University of Southern Denmark, Odense, Denmark; 6Liva Healthcare, Copenhagen, DenmarkNutrients2022;14:3424This study investigated whether LIVA 2.0 (long‐term Lifestyle change InterVention and eHealth Application), a Danish eHealth lifestyle coaching program, could influence significant weight loss and decreased hemoglobin A1c (HbA1c) in patients with type 2 diabetes (T2D).MethodsFrom 2018 to 2019, this randomized controlled trial enrolled through advertising, social media, and clinical contacts 170 participants. The study's inclusion criteria were T2D diagnosis, body mass index (BMI) 30−45 kg/m2, and age 18−70 years. The exclusion criteria were lack of internet access, pregnancy or planning a pregnancy, or a serious disease. The participants were randomized into two groups: 100 (49 women) in the intervention group, who received digital/virtual lifestyle coaching, and 70 (32 women) in the control group who received standard care. The mean age of patients was 56 years, mean HbA1c was 7.4%, and mean body weight was 104 kg. The primary and secondary outcomes were reductions in body weight and HbA1c.ResultsAt six months, 75 patients (75%) in the intervention group and 53 patients (76%) in the control group remained in the trial. The mean body weight loss was 4.2 kg (95% CI, −5.49 to −2.98) in the intervention group and 1.5 kg (95% CI, −2.57 to −0.48) in the control group (P = 0.005). In the intervention group, 24 out of 62 patients with elevated HbA1c at baseline (39%) had a normalized HbA1c < 6.5% at 6 months, compared with eight out of 40 patients with elevated HbA1c at baseline (20%) in the control group (P = 0.047).ConclusionsThe LIVA 2.0 eHealth lifestyle coaching program can lead to significant weight loss and decreased HbA1c among patients with T2D compared with standard care.Use of Video‐based Telehealth Services Using a Mobile App for Workers in Underserved Areas during the COVID‐19 Pandemic: A Prospective Observational StudyPark HS1,2, Jeong S2, Chung HY2,3, Soh JY4, Hyun YH1, Bang SH1, Kim HS41Digital Healthcare Department, BIT Computer Co. Ltd., Seoul, Republic of Korea; 2Department of Medical Informatics, Kyungpook National University, Daegu, Republic of Korea; 3Department of Surgery, Kyungpook National University Hospital, Daegu, Republic of Korea; 4Elecmarvels Co. Ltd., Daegu, Republic of KoreaInt J Med Inform2022;166:104844The limitations on face‐to‐face treatment during the COVID‐19 pandemic have resulted in changes to the structure of existing health‐care services. This study investigated the effects of video‐based telehealth services using a mobile personal health record (PHR) app among industrial workers with metabolic risk factors who live in underserved areas and have relatively poor access to health care.MethodsThe participants in this 16‐week prospective observational study included 117 workers and 27 health‐care professionals. The workers visited their occupational health center three times (at weeks 1, 8, and 16) to undergo health check‐ups and used various features of the mobile PHR app. The health‐care professionals observed the participants' data using the monitoring system and performed appropriate interventions via text‐based and video consultations. The primary outcome was the measured changes to the participants' metabolic risk factors; the secondary outcomes were changes in the participants' lifestyles and their satisfaction with the service as well as observations gleaned from usage logs. One‐way repeated measures analysis of variance and Scheffé's test were performed to observe changes in the participants' health status and lifestyle; a paired t test was performed to analyze changes in service satisfaction. Finally, semistructured questionnaires and in‐depth interviews with the health‐care professionals provided their perspectives after the end of the study.ResultsSignificant improvements at each time point were seen in systolic blood pressure (F = 7.32, P < 0.001), diastolic blood pressure (F = 11.30, P < 0.001), body weight (F = 29.53, P < 0.001), body mass index (F = 17.31, P < 0.001), waist circumference (F = 17.33, P < 0.001), fasting blood glucose (F = 5.11, P = 0.007), and triglycerides (F = 4.66, P = 0.01). No improvements were found in high‐density lipoprotein cholesterol (F = 3.35, P = 0.067). The dietary score (F = 3.26, P = 0.04) showed a significant improvement with each time point, but physical activity (F = 1.06, P = 0.34) did not. Only lifestyle improvement (P < 0.001) showed a significant difference in terms of service satisfaction. The health-care professionals all felt that COVID‐19 had affected their performance, and they also felt the need for the expanded health-care technology services.ConclusionsVideo‐based telehealth services can be an effective means of supporting health and lifestyle interventions for workers, and the results support their use in the occupational health and safety fields.CommentsObesity has been increasing and effecting more individuals over the years. Together with the increase in obesity has been an increase in comorbidities such as T2D, stroke, and cardiovascular disease. These three studies illustrate the benefits of virtual care while dealing with obesity associated with or without T2D and impaired glucose regulation (IGR).Fraticelli and colleagues examined obesity and T2D in Italian patients before and after the COVID‐19 lockdown. The sample size was small, but it was a properly designed study that lasted 6 months. The 36 participants were assigned to either the traditional arm (face‐to‐face contacts) or the web‐based nutritional arm. The study concluded that the web‐based nutritional intervention resulted in a decrease in both body weight and body mass index (BMI) and was effective for both patients with T2D and IGR.Similarly, Christensen and colleagues investigated the use of virtual care in the form of an eHealth lifestyle coaching program for patients with T2D and obesity. This randomized controlled trial (RCT) in Denmark included 170 patients for one year (2018–2019). The study showed a mean body weight loss of 4.2 kg for the intervention group compared with 1.5 kg for the control group. Furthermore, a higher percentage of patients in the intervention group were able to maintain their HbA1c levels of < 6.5% compared with the control group (39% and 20%, respectively) at 6 months.Park and colleagues investigated the use of video‐based telehealth via a mobile app for workers who reside in underserved areas. Their study results demonstrated that there were improvements in blood pressure, weight, waist circumference, fasting blood glucose, and triglycerides. The use of virtual care in the form of a mobile app or telehealth visit provided access to obesity treatment for patients who had barriers, such as being unable to take time off from their work schedule for traditional visits.These three studies highlight that virtual treatment can result in improvements in weight and BMI for obese patients with associated T2D or IGR. These benefits can also extend to improvements in blood pressure and glucose control measured by HbA1c.DIABETIC RETINOPATHY AND TELEHEALTHFindings from a Statewide TeleRetina Diabetic Retinopathy Screening Program in ArkansasShirey M1, Kwok A1, Jenkins H2, Uwaydat S11University of Arkansas for Medical Sciences: Harvey and Bernice Jones Eye Institute, Little Rock, AR; 2University of Arkansas for Medical Sciences, Little Rock, ARInt J Telemed Appl2023;2023:3233803A significant proportion of diabetic patients in the United States do not present for annual dilated eye examinations to monitor for signs of diabetic retinopathy (DR). This study analyzed the results of a statewide, multiclinic teleretinal program to screen rural patients.MethodsAt 10 primary care clinics across Arkansas, patients with diabetes were offered teleretinal imaging services. The images were transmitted to the Harvey and Bernice Jones Eye Institute (JEI) at the University of Arkansas for Medical Sciences for grading and recommendations for further treatment.ResultsFrom February 2019 to May 2022, 668 patients underwent retinal imaging, and 645 images were deemed of sufficient quality to generate an interpretation. No evidence of DR was found in 541 patients; 104 patients had some evidence of DR. No evidence of maculopathy was found in 587 patients; 58 patients had some evidence of maculopathy. Other pathologies were found in 246 patients, with the most common being hypertensive retinopathy, suspected glaucoma, and cataracts.ConclusionsThe JEI TeleRetina program successfully identified DR and other nondiabetic ocular pathologies in rural primary care settings, allowing for appropriate triage.Effectiveness of Telemedicine Diabetic Retinopathy Screening in the USA: A Protocol for Systematic Review and Meta‐analysisPadilla Conde T1, Robinson L2, Vora P1, Ware SL1, Stromberg A3, Bastos de Carvalho A11Department of Ophthalmology and Visual Science, University of Kentucky, Lexington, KY; 2Medical Center Library, University of Kentucky, Lexington, KY; 3Department of Statistics, University of Kentucky, Lexington, KYSyst Rev2023;12:48Diabetic retinopathy (DR) is the leading cause of vision loss among adults in the United States, but vision loss associated with diabetic retinopathy can be prevented with annual retinal examinations leading to timely ophthalmologic care. A systematic review could collate evidence from existing studies to investigate whether using telemedicine to screen for DR in U.S. primary care clinics leads to increased DR screening rates.MethodsRelevant studies will be identified through searching Medline/PubMed interface, Scopus, and Web of Science from their inception until November 2021, as well as searching the reference lists of included studies and previous related review articles or systematic reviews with no restrictions on study design. The eligible studies will include U.S. participants with either type 1 or type 2 diabetes, use telemedicine technology to screen for DR, provide screening rates or the data necessary to calculate such rates. Two reviewers will evaluate the search results independently and make risk‐of‐bias assessments and data extractions. The Cochrane risk‐of‐bias tool (RoB 2) and the Newcastle‐Ottawa scale (NOS) tools will be used to assess the quality and validity of individual studies. Random‐effects meta‐analysis and subgroup analyses will explore potential heterogeneity sources (setting, socioeconomic status, age, ethnicity, study design, outcomes).ConclusionsThe methods for systematic review and synthesis of evidence on telemedicine DR screening and its effect on DR screening rates are outlined in this protocol. The results will be useful for policy‐makers and program managers tasked with designing and implementing evidence‐based services to prevent and manage diabetes and its complications in similar settings.Tele‐ophthalmology for Age‐related Macular Degeneration during the COVID‐19 Pandemic and BeyondMintz J1, Labiste C1, DiCaro MV2, McElroy E1, Alizadeh R3, Xu K31Dr. Kiran C. Patel College of Allopathic Medicine, Nova Southeastern University, Davie, FL; 2College of Medicine, University of Arizona, Tucson, AZ; 3Department of Ophthalmology, University of Arizona, Tucson, AZJ Telemed Telecare2022;28:670–679The care of individuals with age‐related macular degeneration (AMD) has been disrupted by the COVID‐19 pandemic. As a result, many practices are using forms of tele‐ophthalmology services for their patients, and additional guidance is needed on how to maintain continuity of care.MethodsTo identify AMD outcomes and telecare management strategies that could be used during the COVID‐19 pandemic, a literature search was conducted up to August 1, 2020, using Google Scholar, Medline, Web of Science, Pubmed, and LitCovid (search terms: age related macular degeneration, telemedicine, tele‐ophthalmology, COVID‐19, SARS‐COV‐2, retina, anti‐VEGF injections, home monitoring, neovascular, artificial intelligence [AI], fundus photography, fundus angiography, optical coherence tomography, age related eye disease study, home monitoring, primary care eye imaging, Topcon Maestro2, and smartphone). Four American Academy of Ophthalmology and Centers for Disease Control and Prevention web resources were also included.ResultsA 237 total of articles were retrieved; after excluding duplicates, editorials, studies missing patient outcomes, conference abstracts, retracted studies, and case reports, 56 studies were selected for analysis.ConclusionsRisk‐stratification models, developed to readily screen existing patients for their future risk of neovascular AMD (NAMD), can be used with at‐home monitoring devices to detect NAMD and allow providers to determine who should be contacted via tele‐ophthalmology for screening. Telemedicine triage also can be used for new complaints of vision loss to determine who should be referred to a retinal specialist for management of suspected NAMD. Smartphone fundus photography images, sent to a centralized tele‐ophthalmology service, can aid in NAMD detection. Tele‐ophthalmology also can screen patients for the presence of COVID‐19 before in‐person office visits and connect remotely with nursing home, rural, and socioeconomically disadvantaged patients.Implementation of Deep Learning Artificial Intelligence in Vision‐threatening Disease Screenings for an Underserved Community during COVID‐19Zhu A, Tailor P, Verma R, Zhang I, Schott B, Ye C, Szirth B, Habiel M, Khouri ASInstitute of Ophthalmology & Visual Science, Rutgers New Jersey Medical School, Newark, NJJ Telemed Telecare2023;1357633X231158832.Age‐related macular degeneration, diabetic retinopathy, and glaucoma are leading causes of vision loss, and many studies have validated deep‐learning artificial intelligence (AI) for image‐based diagnosis of such vision‐threatening diseases. Our study prospectively investigated deep‐learning AI applications in student‐run non‐mydriatic screenings for an underserved, primarily Hispanic community during COVID‐19.MethodsFive supervised, student‐run community screenings were held in West New York, NJ. The medical students led the clinic participants in non‐mydriatic 45‐degree retinal imaging, and the images were uploaded to a cloud‐based deep‐learning AI for vision‐threatening disease referral. An on‐site tele‐ophthalmology grader and remote clinical ophthalmologist graded the images, with adjudication by a senior ophthalmologist to establish the gold standard diagnosis, which was used to assess the performance of deep‐learning AI.ResultsA total of 385 eyes from 195 screening participants were included (mean age: 52.43 ± 14.5 years; 40.0% female), and a total of 48 participants were referred for at least one vision‐threatening disease. The deep‐learning AI marked 150 of 385 (38.9%) eyes as ungradable, compared with 10 of 385 (2.6%) considered ungradable as per the human gold standard (P < 0.001). Deep‐learning AI had 63.2% sensitivity, 94.5% specificity, 32.0% positive predictive value, and 98.4% negative predictive value in vision‐threatening disease referrals. Deep‐learning AI successfully identified all four eyes with multiple vision‐threatening diseases. Deep‐learning AI graded images (35.6 ± 13.3 seconds) faster than the tele‐ophthalmology grader (129 ± 41.0 seconds) and clinical ophthalmologist (68 ± 21.9 seconds, P < 0.001).ConclusionsIn underserved communities, deep‐learning AI can increase the efficiency and accessibility of vision‐threatening disease screenings and should be adaptable to a variety of environments. In the form of computer‐aided or autonomous diagnosis, deep‐learning AI has potential real‐world applications.CommentsThe leading cause of vision loss among adults in the United States is diabetic retinopathy (DR). Because routine ophthalmologic care can prevent advanced DR and its associated vision loss, an annual retinal examination is recommended for individuals with diabetes. Telehealth can help to overcome the challenges patients face in accessing this care.Shirey and colleagues analyzed the results of a multiclinic teleretinal program designed to screen patients in rural Arkansas in the United States. In this study, 668 patients underwent retinal imaging, and 645 of the images obtained were of good quality for interpretation. No DR was found in 541 patients, amounting to nearly 80% of the cohort. Some evidence of DR was found in 104 patients, evidence of maculopathy in 58 patients, and other pathologies in 246 patients, most commonly hypertensive retinopathy, glaucoma suspects, and cataracts.Padilla Conde and colleagues provided a meta‐analysis and discussion of the need for guidelines in implementing and assessing telehealth for DR. They propose a protocol for reviewing and synthesizing the evidence from telemedicine on rates of DR. The results should help provide evidence‐based services in preventing DR and other associated diabetes‐related complications.The role of tele‐ophthalmology in screening for age‐related macular degeneration during the COVID‐19 pandemic was examined in the study by Mintz and colleagues. The results of this meta‐analysis of 56 studies support the advantages of using tele‐ophthalmology in future to connect with socioeconomically disadvantaged patients, especially those in rural areas and nursing homes.The introduction of artificial intelligence (AI) in medicine can potentially assist in diagnosis and health‐care accessibility. The article by Zhu and colleagues discusses the role of deep learning and AI in computer‐aided diagnosis of eye disease. In their study of a screening clinic, 150 of 385 eyes (38.9%) were marked as ungradable by AI, compared with 10 of 385 (2.6%) graded by the gold standard practice of human intervention (P < 0.001). In vision‐threatening disease referrals, deep‐learning AI had 63.2% sensitivity, 94.5% specificity, 32.0% positive predictive value, and 98.4% negative predictive value. This study reinforces the real‐world applications of deep‐learning AI and its role in screening for DR. The use of virtual care for patients with DR and macular degeneration may be safe and avoids the potential barrier of travel to an office.PREGNANCY AND TELEHEALTHTelehealth Adoption during COVID‐19: Lessons Learned from Obstetric Providers in the Rocky Mountain WestHolman C1, Glover A1,2, McKay K2, Gerard C21Rural Institute for Inclusive Communities and 2School of Public and Community Health Sciences, University of Montana, Missoula, MTTelemed Rep2023;4:1–9Although obstetric providers had used telemedicine to manage gestational diabetes, mental health, and prenatal care, the uptake of telemedicine in the field was less than enthusiastic until the advent of the COVID‐19 pandemic. This study examined the experience of adapting to telehealth among obstetric providers in the Rocky Mountain West to identify implications for policy and practice.MethodsTwenty semistructured interviews were conducted with obstetric providers in Montana, Idaho, and Wyoming following a moderator's guide based on the Aday & Andersen Framework for the Study of Access to Medical Care (1) to explore the domains of health policy, the health system, the utilization of health services, and the population at risk. All the interviews were recorded, transcribed, and analyzed using thematic analysis.ResultsThe providers viewed telehealth as a useful tool during prenatal and postpartum care, and many planned to continue telehealth practices after the pandemic. They shared that their patients reported benefits to telehealth beyond COVID‐19 safety—including limiting their travel time, reducing their time away from work, and alleviating their childcare needs. The providers were concer