American Journal of Preventive Medicine, Journal Year: 2023, Volume and Issue: 66(3), P. 568 - 572
Published: Dec. 4, 2023
Language: Английский
Clinical Orthopaedics and Related Research, Journal Year: 2023, Volume and Issue: 481(8), P. 1623 - 1630
Published: May 23, 2023
Abstract Background Advances in neural networks, deep learning, and artificial intelligence (AI) have progressed recently. Previous learning AI has been structured around domain-specific areas that are trained on dataset-specific of interest yield high accuracy precision. A new model using large language models (LLM) nonspecific domain areas, ChatGPT (OpenAI), gained attention. Although demonstrated proficiency managing vast amounts data, implementation knowledge remains a challenge. Questions/purposes (1) What percentage Orthopaedic In-Training Examination questions can generative, pretrained transformer chatbot (ChatGPT) answer correctly? (2) How does compare with results achieved by orthopaedic residents different levels, if scoring lower than the 10th percentile relative to 5th-year is likely correspond failing American Board Surgery score, this LLM pass surgery written boards? (3) Does increasing question taxonomy affect LLM’s ability select correct choices? Methods This study randomly selected 400 3840 publicly available based compared mean score who took test over 5-year period. Questions figures, diagrams, or charts were excluded, including five could not provide an for, resulting 207 administered raw recorded. The ranking residents. Based findings earlier study, pass-fail cutoff was set at percentile. answered then categorized Buckwalter recall, which deals increasingly complex levels interpretation application knowledge; comparison made performance across taxonomic analyzed chi-square test. Results 47% (97 207) time, 53% (110 time it incorrectly. prior testing, scored 40th for postgraduate year (PGY) 1s, eighth PGY2s, first PGY3s, PGY4s, PGY5s; latter finding (and predefined PGY5s as threshold passing score), seems unlikely would board examination. decreased level increased (it 54% [54 101] Tax 1 correctly, 51% [18 35] 2 34% [24 71] 3 correctly; p = 0.034). Conclusion general-domain low likelihood examination, testing comparable first-year resident. LLM's accurate answers declines complexity, indicating deficiency implementing knowledge. Clinical Relevance Current appears perform better interpretation-based inquires, other opportunity, may become additional tool education.
Language: Английский
Citations
111
Healthcare, Journal Year: 2024, Volume and Issue: 12(3), P. 300 - 300
Published: Jan. 24, 2024
The remarkable progress in data aggregation and deep learning algorithms has positioned artificial intelligence (AI) machine (ML) to revolutionize the field of medicine. AI is becoming more prevalent healthcare sector, its impact on orthopedic surgery already evident several fields. This review aims examine literature that explores comprehensive clinical relevance AI-based tools utilized before, during, after anterior cruciate ligament (ACL) reconstruction. focuses current applications future prospects preoperative management, encompassing risk prediction diagnostics; intraoperative tools, specifically navigation, identifying complex anatomic landmarks during surgery; postoperative terms care rehabilitation. Additionally, educational training settings are presented. Orthopedic surgeons showing a growing interest AI, as evidenced by discussed this review, particularly those related ACL injury. exponential increase studies applicable management tears promises significant application, with attention from surgeons.
Language: Английский
Citations
18
Cureus, Journal Year: 2024, Volume and Issue: unknown
Published: April 24, 2024
This study aims to compare the performance of ChatGPT-3.5 (GPT-3.5) and ChatGPT-4 (GPT-4) on American Society for Surgery Hand (ASSH) Self-Assessment Examination (SAE) determine their potential as educational tools.
Language: Английский
Citations
9
Contemporary Drug Problems, Journal Year: 2022, Volume and Issue: 50(1), P. 3 - 24
Published: Sept. 30, 2022
As harm reduction programs and services proliferate, people who use drugs (PWUD) are increasingly subjected to surveillance through the collection of their personal information, systematic observation, other means. The data generated from these practices frequently repurposed across various institutional sites for clinical, evaluative, epidemiological, administrative uses. Rationales provided increased include more effective provision care, service optimization, risk stratification, efficiency in resource allocation. With this mind, our reflective essay draws on empirical analysis work within movements reflect critically impacts implications expansion. While we argue that many not inherently problematic or harmful, unchecked expansion under a banner health may contribute decreased uptake services, rationing conditionalities tied access, potential deepening disparities amongst some PWUD, an overlay criminal-legal systems. In context, relies enlistment range therapeutic actors reflects permeable boundary between care control. We thus call broader critical dialogue problems posed by settings, end sharing information with law enforcement criminal legal actors, deference stated need among PWUD meaningful anonymity when accessing services.
Language: Английский
Citations
28
Translational Medicine UniSa, Journal Year: 2024, Volume and Issue: 26(1)
Published: Feb. 25, 2024
Aims: This study delves into the two-year opioid prescription trends in Local Sanitary Agency Naples 3 South, Campania Region, Italy. The research aims to elucidate prescribing patterns, demographics, and dosage categories within a population representing 1.7% of national total. Perspectives on artificial intelligence are discussed.
Language: Английский
Citations
7
Epidemiology, Journal Year: 2024, Volume and Issue: 35(2), P. 232 - 240
Published: Jan. 2, 2024
Background: Drug overdose persists as a leading cause of death in the United States, but resources to address it remain limited. As result, health authorities must consider where allocate scarce within their jurisdictions. Machine learning offers strategy identify areas with increased future risk proactively prevention resources. This modeling study is embedded randomized trial measure effect proactive resource allocation on statewide rates Rhode Island (RI). Methods: We used data from RI 2016 2020 develop an ensemble machine model predicting neighborhood-level fatal risk. Our integrated gradient boosting and super learner base models moving window framework make predictions 6-month intervals. performance target, developed priori Department Health, was 20% neighborhoods containing at least 40% deaths, including one neighborhood per municipality. The validated after launch. Results: selected priority capturing 40.2% deaths during test periods 44.1% validation periods. outperformed performed comparably best-performing Conclusions: demonstrated capacity for predict degree accuracy suitable practitioners. Jurisdictions may predictive tool guide
Language: Английский
Citations
5
Cureus, Journal Year: 2024, Volume and Issue: unknown
Published: March 13, 2024
Introduction Artificial intelligence (AI) models using large language (LLMs) and non-specific domains have gained attention for their innovative information processing. As AI advances, it's essential to regularly evaluate these tools' competency maintain high standards, prevent errors or biases, avoid flawed reasoning misinformation that could harm patients spread inaccuracies. Our study aimed determine the performance of Chat Generative Pre-trained Transformer (ChatGPT) by OpenAI Google BARD (BARD) in orthopedic surgery, assess based on question types, contrast between different AIs compare residents. Methods We administered ChatGPT 757 Orthopedic In-Training Examination (OITE) questions. After excluding image-related questions, answered 390 multiple choice all categorized within 10 sub-specialties (basic science, trauma, sports medicine, spine, hip knee, pediatrics, oncology, shoulder elbow, hand, food ankle) three taxonomy classes (recall, interpretation, application knowledge). Statistical analysis was performed analyze number questions correctly each model, returned model sub-specialty designation, comparison results residents classified respective post-graduate year (PGY) level. Results more overall (58% vs 54%, p<0.001). better medicine basic science worse hand while (p<0.05). The recall compared knowledge Based previous data, it ranked 42nd-96th percentile ones (PGY1s), 27th-58th PGY2s, 3rd-29th PGY3s, 1st-21st PGY4s, 1st-17th PGY5s. Discussion excelled but fell short both well poorly knowledge-based than overall. Although reached second-year PGY resident level, passing American Board Surgery (ABOS). Its strengths recall-based inquiries highlight its potential as an learning educational tool.
Language: Английский
Citations
5
Current Opinion in Psychology, Journal Year: 2025, Volume and Issue: 62, P. 102003 - 102003
Published: Jan. 30, 2025
Language: Английский
Citations
0
Cureus, Journal Year: 2025, Volume and Issue: unknown
Published: March 29, 2025
Background: Opioid use disorder (OUD) is associated with significantly increased mortality rates compared to the general population, driven by overdose risk, high-risk behaviors, and comorbid conditions. While opioid agonist treatment reduces mortality, identifying risk factors for death among individuals OUD remains critical improving outcomes. Methods: A retrospective analysis of 2020 National Readmission Database identified admissions using International Classification Diseases, 10th Revision, Clinical Modification codes. Patients over 18 years age were included, statistical analyses, including logistic regression, assessed 30-day readmission predictors. Data analyzed IBM Statistical Package Social Sciences Statistics Windows, version 1.0.0.1327 (IBM Corp., Armonk, NY). Results: Nonsurvivors generally older (median age: 58 vs. 47 years) had a higher prevalence severe comorbidities, cardiac arrest (24.1% 0.4%, p < 0.001), respiratory failure (83.0% 16.2%, acute kidney injury (61.9% 16.8%, 0.001). Mortality was more common patients Medicare (44.4% 31.7%) in larger hospitals. Psychiatric conditions, such as depression suicidal ideation, frequent survivors, suggesting potential protective effects or earlier intervention. Multivariable (odds ratio, OR: 20.210, (OR: 9.993, liver 4.298, 0.001) strongest predictors, while female sex psychiatric disorders lower risk. Conclusion: influenced age, hospital characteristics, healthcare disparities. Integrated care approaches that address both medical conditions are essential survival Future research should focus on targeted interventions mitigate enhance harm reduction strategies this vulnerable population.
Language: Английский
Citations
0
Scientifica, Journal Year: 2025, Volume and Issue: 2025(1)
Published: Jan. 1, 2025
Background: Pain is a significant symptom in cancer patients that frequently not effectively treated, and managing it seen as crucial aspect of caring for these patients. This severe pain causes disturbance their quality life. At present, there are different challenges utilizing range pharmacological nonpharmacological treatments Recent technological advancements, particularly artificial intelligence, have improved the management Artificial intelligence its algorithms offer potential solutions relief with reduced side effects. Study Design: The current review aimed to assess validity studies on using Four databases been used all published from start 2023: PubMed, Scopus, Web Science, Google Scholar. search mechanism articles was mainly valid mesh‐based keywords, asking experts, reviewing literature including “Pain,” “Pain management,” “Cancer,” “Artificial intelligence.” During initial search, total 450 were found, after considering inclusion exclusion criteria abstract content articles, 15 finally included study. Results: AI‐based can provide individual plans. When AI analyzes large patient data such physiological signals, responses treatment, symptoms who diagnosed pain, possible accurately adjust therapeutic measures. Conclusions: enables healthcare providers timely care assistance through remote monitoring telehealth services, even when they physically present. Despite presence hurdles ensuring ethical practices protecting privacy, integration oncology brings optimism future.
Language: Английский
Citations
0