Search

PURPOSE: Automated machine learning (AutoML) has emerged as a novel tool for medical professionals lacking coding experience, enabling them to develop predictive models for treatment outcomes. This study evaluated the performance of AutoML tools in developing models predicting the success of pneumatic retinopexy (PR) in treatment of rhegmatogenous retinal detachment (RRD). These models were then compared with custom models created by machine learning (ML) experts. DESIGN: Retrospective multicenter study. PARTICIPANTS: Five hundred and thirty nine consecutive patients with primary RRD that underwent PR by a vitreoretinal fellow at 6 training hospitals between 2002 and 2022. METHODS: We used 2 AutoML platforms: MATLAB Classification Learner and Google Cloud AutoML. Additional models were developed by computer scientists. We included patient demographics and baseline characteristics, including lens and macula status, RRD size, number and location of breaks, presence of vitreous hemorrhage and lattice degeneration, and physicians experience. The dataset was split into a training (n = 483) and test set (n = 56). The training set, with a 2:1 success-to-failure ratio, was used to train the MATLAB models. Because Google Cloud AutoML requires a minimum of 1000 samples, the training set was tripled to create a new set with 1449 datapoints. Additionally, balanced datasets with a 1:1 success-to-failure ratio were created using Python. MAIN OUTCOME MEASURES: Single-procedure anatomic success rate, as predicted by the ML models. F2 scores and area under the receiver operating curve (AUROC) were used as primary metrics to compare models. RESULTS: The best performing AutoML model (F2 score: 0.85; AUROC: 0.90; MATLAB), showed comparable performance to the custom model (0.92, 0.86) when trained on the balanced datasets. However, training the AutoML model with imbalanced data yielded misleadingly high AUROC (0.81) despite low F2-score (0.2) and sensitivity (0.17). CONCLUSIONS: We demonstrated the feasibility of using AutoML as an accessible tool for medical professionals to develop models from clinical data. Such models can ultimately aid in the clinical decision-making, contributing to better patient outcomes. However, outcomes can be misleading or unreliable if used naively. Limitations exist, particularly if datasets contain missing variables or are highly imbalanced. Proper model selection and data preprocessing can improve the reliability of AutoML tools. FINANCIAL DISCLOSURES: Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article.

Over the past decade behavioral science, particularly psychology, has come under fire from critics for being fixated on progressive political ideology, most notably Diversity, Equity, and Inclusion (DEI). The critics' [...]

Purpose: Automated machine learning (AutoML) has emerged as a novel tool for medical professionals lacking coding experience, enabling them to develop predictive models for treatment outcomes. This study evaluated the performance of AutoML tools in developing models predicting the success of pneumatic retinopexy (PR) in treatment of rhegmatogenous retinal detachment (RRD). These models were then compared with custom models created by machine learning (ML) experts. Design: Retrospective multicenter study. Participants: Five hundred and thirty nine consecutive patients with primary RRD that underwent PR by a vitreoretinal fellow at 6 training hospitals between 2002 and 2022. Methods: We used 2 AutoML platforms: MATLAB Classification Learner and Google Cloud AutoML. Additional models were developed by computer scientists. We included patient demographics and baseline characteristics, including lens and macula status, RRD size, number and location of breaks, presence of vitreous hemorrhage and lattice degeneration, and physicians’ experience. The dataset was split into a training (n = 483) and test set (n = 56). The training set, with a 2:1 success-to-failure ratio, was used to train the MATLAB models. Because Google Cloud AutoML requires a minimum of 1000 samples, the training set was tripled to create a new set with 1449 datapoints. Additionally, balanced datasets with a 1:1 success-to-failure ratio were created using Python. Main Outcome Measures: Single-procedure anatomic success rate, as predicted by the ML models. F2 scores and area under the receiver operating curve (AUROC) were used as primary metrics to compare models. Results: The best performing AutoML model (F2 score: 0.85; AUROC: 0.90; MATLAB), showed comparable performance to the custom model (0.92, 0.86) when trained on the balanced datasets. However, training the AutoML model with imbalanced data yielded misleadingly high AUROC (0.81) despite low F2-score (0.2) and sensitivity (0.17). Conclusions: We demonstrated the feasibility of using AutoML as an accessible tool for medical professionals to develop models from clinical data. Such models can ultimately aid in the clinical decision-making, contributing to better patient outcomes. However, outcomes can be misleading or unreliable if used naively. Limitations exist, particularly if datasets contain missing variables or are highly imbalanced. Proper model selection and data preprocessing can improve the reliability of AutoML tools. Financial Disclosure(s): Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article.

Abstract Background In 2020 COVID-19 was the third leading cause of death in the United States. Increases in suicides, overdoses, and alcohol related deaths were seen—which make up deaths of despair. How deaths of despair compare to COVID-19 across racial, ethnic, and gender subpopulations is relatively unknown. Preliminary studies showed inequalities in COVID-19 mortality for Black and Hispanic Americans in the pandemic's onset. This study analyzes the racial, ethnic and gender disparities in years of life lost due to COVID-19 and deaths of despair (suicide, overdose, and alcohol deaths) in 2020. Methods This cross-sectional study calculated and compared years of life lost (YLL) due to Deaths of Despair and COVID-19 by gender, race, and ethnicity. YLL was calculated using the CDC WONDER database to pull death records based on ICD-10 codes and the Social Security Administration Period Life Table was used to get estimated life expectancy for each subpopulation. Results In 2020, COVID-19 caused 350,831 deaths and 4,405,699 YLL. By contrast, deaths of despair contributed to 178,598 deaths and 6,045,819 YLL. Men had more deaths and YLL than women due to COVID-19 and deaths of despair. Among White Americans and more than one race identification both had greater burden of deaths of despair YLL than COVID-19 YLL. However, for all other racial categories (Native American/Alaskan Native, Asian, Black/African American, Native Hawaiian/Pacific Islander) COVID-19 caused more YLL than deaths of despair. Also, Hispanic or Latino persons had disproportionately higher mortality across all causes: COVID-19 and all deaths of despair causes. Conclusions This study found greater deaths of despair mortality burden and differences in burden across gender, race, and ethnicity in 2020. The results indicate the need to bolster behavioral health research, support mental health workforce development and education, increase access to evidence-based substance use treatment, and address systemic inequities and social determinants of deaths of despair and COVID-19.

PurposeTo report a case of visual recovery and vascular reperfusion after vaso-occlusive retinopathy from anti-phospholipid syndrome associated with systemic lupus erythematosus.ObservationsA 15-year-old boy with a known diagnosis of systemic lupus erythematosus and a clinically significant anti-phospholipid panel presented with sudden vision loss in the left eye. Examination and ocular imaging revealed signs of vaso-occlusive retinopathy. The patient was immediately started on high dose intravenous steroids, followed by mycophenolate mofetil. He remained on aspirin. After showing no improvement in retinal arteriole and capillary perfusion he was started on therapeutic anti-coagulation with enoxaparin. He regained 20/20 vision. Intravenous fluorescein angiography demonstrated reperfusion of retinal arterioles. Optical coherence tomography angiography showed return of flow in the capillary networks.ConclusionsWe present a case of vaso-occlusive retinopathy in a patient with known systemic lupus erythematosus and a clinically significant anti-phospholipid panel, thus meeting criteria for anti-phospholipid syndrome. He was treated with intravenous methylprednisolone, mycophenolate motefil, aspirin, and enoxaparin. The patient not only had great recovery of visual acuity, but also demonstrated reperfusion of arterioles and reconstitution of flow in the retinal capillary network. These findings suggest that the vaso-occlusive disease is reversible if the diagnosis is made promptly and intensive therapy is initiated.ImportanceCurrently there are no reported cases of vaso-occlusive retinopathy from APLS and SLE with visual recovery, reperfusion, and return of capillary flow.

Background Intravenous thrombolysis (IVT) and endovascular therapy (EVT) are both important treatments for large‐vessel occlusion stroke. However, it is still unclear how the timing of IVT, EVT, and the need for transfer of a patient to an endovascular stroke center for EVT affect outcomes. In this investigation, we study the interaction between IVT, rapidity to EVT, and need for transfer among patients with large‐vessel occlusion stroke. Methods This investigation is an analysis of the OPUS‐REACH (Optimizing the Use Prehospital Stroke Systems of Care–Reacting to Changing Paradigms) registry of patients with large‐vessel occlusion stroke from 9 endovascular centers in the United States. Using the database, we extracted baseline characteristics of patients, whether the patient received IVT, and time intervals in the patients’ care. Patient demographics and characteristics were compared between 2 groups using the χ2 test for categorical variables and 2‐sample t‐tests or Wilcoxon rank‐sum tests for continuous variables. Multivariable logistic regression was performed to determine the adjusted associations of the variables with 90‐day dichotomized modified Rankin Scale outcome. Results A total of 1171 patients were included in the final analysis, and 38.9% had good functional outcome at 90 days. Male sex and lower initial National Institutes of Health Stroke Scale score were nonmodifiable factors associated with good clinical outcomes. We saw no differences in outcome whether a patient underwent primary or secondary transport. On multiple variable analysis, the receipt of IVT was the only modifiable factor associated with good outcomes. We found no overall effect of time from last known well to EVT on 90‐day outcomes unless the patient received IVT. Conclusions In this investigation, receipt of IVT was independently associated with improved outcomes at 90 days with an odds ratio of 1.51. Neither shorter time from last known well to EVT nor direct transport to an endovascular stroke center versus transfer to an endovascular stroke center was associated with improved outcomes. We therefore conclude that prehospital algorithms must account for the timely administration of IVT over time to EVT.

ABSTRACTBackground Previous research has found chronic pain to be prevalent among individuals with opioid use disorder (OUD). The perception that pain is related to OUD onset, maintenance, relapse, and treatment delay has been noted in this population. However, prior works primarily involved treatment-engaged populations. Scant research describes such perceptions among non-treatment-seeking individuals.Aims This study describes pain burden and perceptions regarding the role of pain in OUD onset, maintenance, relapse, and addiction treatment delay in a sample of individuals with untreated OUD.Methods This cross-sectional study surveyed syringe exchange participants (n = 141). Participants responded to a survey including Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition OUD criteria, pain survey scales, demographic characteristics, and questions regarding pain and its perceived relatedness to aspects of OUD.Results Most participants reported pain within the past 4 weeks (127, 91.4%). Data displayed a skew toward more intense pain ratings, with 120 reporting their pain as greater than mild (86.3%). A majority of participants agreed that pain was responsible for their OUD onset (79, 56.4%), maintenance (76, 54.3%), past relapse experience (82, 57.9%), and treatment delay (81, 57.9%). Correlative analyses revealed that pain severity and interference measures displayed moderate and statistically significant associations with extent of perceived relatedness of pain to these aspects of OUD.Conclusions Among this sample of individuals with untreated OUD, pain and pain interference were prevalent. Pain was perceived to be related to OUD onset, maintenance, relapse, and treatment delay by a majority of the sample. These findings are in accordance with and expand upon prior works.Abbreviations OUD: opioid use disorder; DSM-5: Diagnostic and Statistical Manual 5; BPI: Brief Pain Inventory; NIDA: National Institute on Drug Abuse; IASP: International Association for the Study of Pain; MOUD: Medications for Opioid Use Disorder; IQR: Interquartile Range.

Most classic genetic approaches utilize binary modifications that preclude the identification of key knockdowns for essential genes or other targets that only require moderate modulation. As a complementary approach to these classic genetic methods, we describe a plasmid-based library methodology that affords bidirectional, graded modulation of gene expression enabled by tiling the promoter regions of all 969 genes that comprise the ito977 model of Saccharomyces cerevisiae's metabolic network. When coupled with a CRISPR-dCas9-based modulation and next-generation sequencing, this method affords a library-based, bidirection titration of gene expression across all major metabolic genes. We utilized this approach in two case studies: growth enrichment on alternative sugars, glycerol and galactose, and chemical overproduction of betaxanthins, leading to the identification of unique gene targets. In particular, we identify essential genes and other targets that were missed by classic genetic approaches.

ObjectiveTo evaluate the outcomes of patients undergoing pneumatic retinopexy (PR) performed by vitreoretinal fellows at 6 academic centers in the United States.DesignRetrospective, multicenter, consecutive case series.Participants483 patients with primary retinal detachments who underwent PR by 49 vitreoretinal fellows from 6 U.S. training sites between 2002 and 2016.MethodsWe reviewed medical records of patients and recorded baseline clinical characteristics (age, sex, baseline visual acuity, lens status, presence of lattice degeneration, presence of vitreous hemorrhage, location of retinal breaks, macular status, and size of detachment), visual and anatomic outcomes at 3-months after PR, as well as training level and PR experience of the fellow at the time of the procedure.Main outcome measuresSingle-procedure anatomic success and visual acuity at 3-months follow-up, and association with clinical and training-related factors.ResultsVitreoretinal fellows performed a variable number of PR, with a median of 7 cases per fellow (range 1-24). Single-procedure anatomic success was 66.8%, and mean LogMAR visual outcome was 0.43 (Snellen equivalent 20/54) at 3 months. Factors that were independently associated with single-procedure success include phakic lens status (P = 0.01), smaller size of retinal detachment (P = 0.02), and the fellow's procedure experience (P = 0.01). The only factor associated with worse visual outcome was baseline visual acuity (P < 0.001).ConclusionVitreoretinal fellows perform variably few pneumatic retinopexies but have outcomes comparable to reported rates by experienced specialists. Procedure experience of individual fellows may impact anatomic outcomes.

Abstract Background Large vessel occlusion (LVO) strokes are best treated with rapid endovascular therapy (EVT). There are two routes that LVO stroke patients can take to EVT therapy when transported by EMS: primary transport (ambulance transports directly to an endovascular stroke center (ESC) or secondary transport (EMS transports to a non-ESC then transfers for EVT). There is no clear evidence which path to care results in better functional outcomes for LVO stroke patients. To find this answer, an analysis of a large, real-world population of LVO stroke patients must be performed. Methods A pragmatic registry of LVO stroke patients from nine health systems across the United States. The nine health systems span urban and rural populations as well as the spectrum of socioeconomic statuses. We will use univariate and multivariate analysis to explore the relationships between type of EMS transport, socioeconomic factors, and LVO stroke outcomes. We will use geographic information systems and spatial analysis to examine the complex movements of patients in time and space. To detect an 8% difference between groups, with a 3:1 patient ratio of primary to secondary transports, 95% confidence and 80% power, we will need approximately 1600 patients. The primary outcome is the patients with modified Rankin Scale (mRS) ≤ 2 at 90 days. Subgroup analyses include patients who receive intravenous thrombolysis and duration of stroke systems. Secondary analyses include socioeconomic factors associated with poor outcomes after LVO stroke. Discussion Using the data obtained from the OPUS-REACH registry, we will develop evidence based algorithms for prehospital transport of LVO stroke patients. Unlike prior research, the OPUS-REACH registry contains patient-level data spanning from EMS dispatch to ninety day functional outcomes. We expect that we will find modifiable factors and socioeconomic disparities associated with poor outcomes in LVO stroke. OPUS-REACH with its breadth of locations, detailed patient records, and multidisciplinary researchers will design the optimal prehospital stroke system of care for LVO stroke patients.

PurposeTo assess public awareness, interest, and concerns regarding Vuity (1.25% pilocarpine hydrochloride ophthalmic solution), an eye drop for the treatment of presbyopia, based on Google Trends.MethodsWe used Google Trends that provides a relative search volume for queried terms, to evaluate searches for Vuity from June 30, 2021, to June 30, 2022, in the United States. The data for this study were downloaded on June 30, 2022. Main outcome measures were changes in relative search volumes for the terms "Vuity," "Eye drops for reading," "Eye drops for near vision," "Presbyopia," "Pilocarpine," and related popular search terms, such as "Vuity side effects," and "Vuity retinal detachment".ResultsSince the approval of Vuity on October 29, 2021, notable increases in the relative search volumes occurred for Vuity in October 2021, December 2021, and from March to April 2022, which coincided with its approval, availability, and subsequent direct-to-consumer advertising based on positive results of clinical trials. The direct-to-consumer advertising had the greatest impact on the search volume for Vuity. Specific interests included Vuity cost, where to buy it, and its side effects. Retinal detachment was the most highly searched Vuity side effect. Geographic variation was evident, with the relative search volumes highest for "Vuity" in Wyoming, followed by North Dakota.ConclusionGoogle Trends is a useful tool for monitoring increases in public interest in Vuity for presbyopia. Public concerns regarding side effects warrant further real-world investigations of the causal relationship between Vuity and retinal detachment.

Objective: To assess global trends in interest surrounding the newly Food and Drug Administration (FDA)-approved treatment for geographic atrophy, (GA), Syfovre (pegcetacoplan), and related searches. Methods: We utilized Google Trends, in order to gauge the public interest in Syfovre from October 16, 2022, to October 8, 2023. Results: Notable spikes in relative search volumes (RSV)s for "Syfovre" were observed in mid-to-late February 2023, and in March and April 2023, coinciding with the drug's FDA approval and introduction to the market. Of the various side effects, retinal vasculitis garnered the most significant attention, with a sharp rise in RSV in mid-July 2023. Geographic variation was evident, with the highest RSVs for "Syfovre" originating from users on the East Coast. Conclusion: Google Trends proves to be a useful tool for gaining insight into public interest in pegcetacoplan as a treatment for GA. [Ophthalmic Surg Lasers Imaging Retina 2024;55:452–460.] [ABSTRACT FROM AUTHOR]

Sex differences in the use of competitive tactics have been well established. Although many factors may contribute to these sex differences, according to social role theory (SRT), stereotypes and expectations about men's and women's typical social roles are crucial. We addressed the potential impact of social roles by studying massively multiplayer online role-playing games (MMORPGs), a setting where individuals represent themselves with avatars and thus enjoy the opportunity to compete without regard to the typical expectations and behaviors associated with men's and women's roles. We surveyed players via MTurk (63 women, 191 men) and Reddit (166 women, 1,326 men) regarding their frequency of engaging in five competitive behaviors and the sex and role of their primary avatar. As expected, there were reliable sex differences in competitiveness: men were more likely than women to engage in player-versus-player duels (MTurk d = 0.19; Reddit d = 0.51), do solo runs of difficult content (0.30, 0.35), and work to acquire expensive items (0.32, 0.19); women were more likely than men to seek in-game awards (−0.38, −0.36) and spend real-world money on expensive microtransactions (−0.16, −0.27). Contrary to SRT, these sex differences in forms of competitive behavior were generally unrelated to players’ chosen avatar sex or avatar role. These results instead indicate that sex differences in competitiveness largely reflect evolved predispositions.

Abstract Introduction Blunt traumatic injuries are a leading cause of morbidity and mortality in the pediatric population. Contrast‐enhanced multidetector computed tomography is the best imaging tool for screening patients at risk of blunt abdominal injury. The Pediatric Emergency Care Applied Research Network (PECARN) abdominal rule was derived to identify patients at low risk for significant abdominal injury who do not require imaging. Methods We conducted a retrospective review of pediatric patients with blunt trauma to validate the PECARN rule in a non‐pediatric specialized hospital from February 3, 2013, through December 31, 2019. We excluded those with penetrating or mild isolated head injury. The PECARN decision rule was retrospectively applied for the presence of a therapeutic intervention, defined as a laparotomy, angiographic embolization, blood transfusion, or administration of intravenous fluids for pancreatic or gastrointestinal injury. Sensitivity and specificity analysis were conducted along with the negative and positive predictive values. Results A total of 794 patients were included in the final analysis; 23 patients met the primary outcome for an acute intervention. The PECARN clinical decision rule (CDR) had a sensitivity of 91.3%, a negative predictive value of 99.5, and a negative likelihood ration of 0.16. Conclusion In a non‐pediatric specialty hospital, the PECARN blunt abdominal CDR performed with comparable sensitivity and negative predictive value to the derivation and external validation study performed at specialized children's hospitals.

Purpose: Orbital cellulitis (OC) is a rare postoperative complication of glaucoma drainage device (GDD) implantation. To date, there have only been 10 reported cases of OC following GDD implantation. Case Report: Here, we report a case of OC in a 57-year-old man who developed pain, proptosis, and limited extraocular motility two days after uneventful Ahmed FP7 implantation in the right eye. Contrast-enhanced computed tomography of the orbits demonstrated fat stranding and a small fluid collection, consistent with OC. He had minimal improvement with intravenous antibiotics and ultimately underwent GDD explantation. A systematic review of the literature showed that the development of OC following GDD implantation can occur in the early or late postoperative period. Immediate hospitalization with intravenous administration of broad-spectrum antibiotics is recommended. Explantation of the infected GDD is often required for source control. Conclusion: OC is a rare postoperative complication of GDD implantation. Prompt evaluation and treatment are required, often combined with GDD explantation.

Purpose: To describe a case of bilateral infectious scleritis secondary to Histoplasma capsulatum in the setting of a locally and systemically immunosuppressed patient. Observations: A 45-year-old man with HLA-B27 associated ankylosing spondylitis and anterior uveitis on systemic secokinumab, underwent bilateral cataract extraction which required extensive peri-operative steroids, including intravitreal triamcinolone, topical prednisolone, and oral prednisone. Six weeks after cataract surgery, the patient presented with mild eye irritation and was found to have bilateral subconjunctival purulence and necrosis. Histoplasma capsulatum was identified on fungal cultures and confirmed by DNA probe. The patient was treated with fortified amphotericin drops and oral itraconazole with complete resolution of scleritis. Conclusion and importance: H. capsulatum is a rare cause of infectious scleritis that must be considered in our immunosuppressed and post-surgical patients.

The purpose of this paper is to share how physical education (PE) offers a unique place of interdisciplinary connections, contributing to coherence in the general education (GE) experience, and to demonstrate the overall value of PE, serving as a rationale for its inclusion in a GE program. A definition of GE is provided, followed by an historical overview of GE in the United States, identifying predominant GE curricular models. An example of how coherence in the GE experience can be obtained through the inclusion of physical education is provided through the examination of a GE model at a California public university.

Aviation fuel (i.e., jet fuel) requires a mixture of C9 -C16 hydrocarbons having both a high energy density and a low freezing point. While jet fuel is currently produced from petroleum, increasing concern with the release of CO2 into the atmosphere from the combustion of petroleum-based fuels has led to policy changes mandating the inclusion of biomass-based fuels into the fuel pool. Here we report a novel way to produce a mixture of branched cyclohexane derivatives in very high yield (>94 %) that match or exceed many required properties of jet fuel. As starting materials, we use a mixture of n-alkyl methyl ketones and their derivatives obtained from biomass. These synthons are condensed into trimers via base-catalyzed aldol condensation and Michael addition. Hydrodeoxygenation of these products yields mixtures of C12 -C21 branched, cyclic alkanes. Using models for predicting the carbon number distribution obtained from a mixture of n-alkyl methyl ketones and for predicting the boiling point distribution of the final mixture of cyclic alkanes, we show that it is possible to define the mixture of synthons that will closely reproduce the distillation curve of traditional jet fuel.