Landau-Kleffner syndrome (LKS) in an 8-year-old girl: a case report and review of the literature.

Introduction and importance: Landau-Kleffner syndrome (LKS) is a rare epileptic encephalopathy characterized by language regression and abnormal electroencephalogram (EEG) patterns. This case report highlights the importance of early recognition and intervention in LKS, as well as the challenges in diagnosis and management due to its varied clinical manifestations. Case presentation: An 8-year-old girl presented with delayed speech, suspected hearing loss, and regression in language skills. Diagnostic tests revealed mild sensorineural hearing loss and EEG abnormalities consistent with LKS. The patient underwent speech therapy and received pharmacological treatment with valproic acid, resulting in significant improvements in language function. Clinical discussion: This case report provides insights into the typical features of LKS, including language regression and EEG abnormalities. It also highlights uncommon findings such as sensorineural hearing loss and mild intellectual delay. The multidisciplinary approach involving neurology, audiology, speech therapy, and education is crucial in the diagnosis and management of LKS. Conclusion: Early recognition and intervention, along with tailored pharmacological approaches and a multidisciplinary care approach, are essential in managing LKS. Further research is needed to better understand the pathophysiology, natural history, and optimal treatment of LKS, aiming to improve long-term outcomes for affected children and their families.

Exploring cross-linguistic differences in parental input and their associations with child expressive language in ASD: Bulgarian versus English comparison.

BACKGROUND: Parental input plays a central role in typical language acquisition and development. In autism spectrum disorder (ASD), characterized by social communicative and language difficulties, parental input presents an important avenue for investigation as a target for intervention. A rich body of literature has identified which aspects of grammatical complexity and lexical diversity are most associated with child language ability in both typical development and autism. Yet, the majority of these studies are conducted with English-speaking children, thus potentially overlooking nuances in parental input derived from cross-linguistic variation. AIMS: To examine the differences in verbal parental input to Bulgarian- and English-speaking children with ASD. To examine whether aspects of verbal parental input found to be concurrent predictors of English-speaking children's expressive language ability are also predictors of the expressive language of Bulgarian-speaking children with ASD. METHODS & PROCEDURES: We compared parental input to Bulgarian-speaking (N = 37; 2;7-9;10 years) and English-speaking (N = 37; 1;8-4;9 years) children with ASD matched on expressive language. Parent-child interactions were collected during free play with developmentally appropriate toys. These interactions were transcribed, and key measures of parental input were extracted. OUTCOMES & RESULTS: English-speaking parents produced more word tokens and word types than Bulgarian-speaking parents. However, Bulgarian parents produced more verbs in relation to nouns and used more statements and exclamations but asked fewer questions than English-speaking parents. In addition, child age and parents' use of questions were significant concurrent predictors of child expressive vocabulary. CONCLUSIONS & IMPLICATIONS: This is one of the first studies to conduct a cross-linguistic comparison of parental input in ASD. The differences found emphasize the need to further study parental input to Bulgarian children and adapt naturalistic parent-mediated interventions to the local language and its specific characteristics. WHAT THIS PAPER ADDS: What is already known on the subject A rich body of literature has identified the specific aspects of grammatical complexity, lexical diversity, and question-asking that are concurrently and longitudinally associated with the language ability of children with typical development and of children with ASD. Yet, the majority of these studies are conducted with English-speaking children. What this paper adds to the existing knowledge The present study finds that there are specific differences in verbal parental input to Bulgarian- and English-speaking children with autism in terms of lexical composition and question-asking. Bulgarian parents used more verbs than nouns, and the opposite pattern was found for English-speaking parents. In addition, Bulgarian parents asked fewer questions but used more statements and exclamations. Nevertheless, parental question use was significantly correlated with children's language ability across both groups, suggesting that question-asking should be further examined as a potential target for parent-mediated language interventions for Bulgarian children with autism. What are the potential or actual clinical implications of this work? Most language and social communication interventions for autism are designed and piloted with English-speaking children. These interventions are often simply translated and used in different countries, with different populations and in different contexts. However, considering that one of the defining characteristics of autism is language difficulty, more studies should examine (1) how these language difficulties manifest in languages other than English, and (2) what characterizes verbal parental input in these other contexts. Such research investigations should inform future language and social communication interventions. The present study emphasizes the cross-linguistic differences between Bulgarian- and English-speaking parents' verbal input to their children with autism.

The Ability of ChatGPT in Paraphrasing Texts and Reducing Plagiarism: A Descriptive Analysis.

Background: The introduction of ChatGPT by OpenAI has garnered significant attention. Among its capabilities, paraphrasing stands out. Objective: This study aims to investigate the satisfactory levels of plagiarism in the paraphrased text produced by this chatbot. Methods: Three texts of varying lengths were presented to ChatGPT. ChatGPT was then instructed to paraphrase the provided texts using five different prompts. In the subsequent stage of the study, the texts were divided into separate paragraphs, and ChatGPT was requested to paraphrase each paragraph individually. Lastly, in the third stage, ChatGPT was asked to paraphrase the texts it had previously generated. Results: The average plagiarism rate in the texts generated by ChatGPT was 45% (SD 10%). ChatGPT exhibited a substantial reduction in plagiarism for the provided texts (mean difference -0.51, 95% CI -0.54 to -0.48; P<.001). Furthermore, when comparing the second attempt with the initial attempt, a significant decrease in the plagiarism rate was observed (mean difference -0.06, 95% CI -0.08 to -0.03; P<.001). The number of paragraphs in the texts demonstrated a noteworthy association with the percentage of plagiarism, with texts consisting of a single paragraph exhibiting the lowest plagiarism rate (P<.001). Conclusions: Although ChatGPT demonstrates a notable reduction of plagiarism within texts, the existing levels of plagiarism remain relatively high. This underscores a crucial caution for researchers when incorporating this chatbot into their work.

Assessing GPT-4's Performance in Delivering Medical Advice: Comparative Analysis With Human Experts.

Background: Accurate medical advice is paramount in ensuring optimal patient care, and misinformation can lead to misguided decisions with potentially detrimental health outcomes. The emergence of large language models (LLMs) such as OpenAI's GPT-4 has spurred interest in their potential health care applications, particularly in automated medical consultation. Yet, rigorous investigations comparing their performance to human experts remain sparse. Objective: This study aims to compare the medical accuracy of GPT-4 with human experts in providing medical advice using real-world user-generated queries, with a specific focus on cardiology. It also sought to analyze the performance of GPT-4 and human experts in specific question categories, including drug or medication information and preliminary diagnoses. Methods: We collected 251 pairs of cardiology-specific questions from general users and answers from human experts via an internet portal. GPT-4 was tasked with generating responses to the same questions. Three independent cardiologists (SL, JHK, and JJC) evaluated the answers provided by both human experts and GPT-4. Using a computer interface, each evaluator compared the pairs and determined which answer was superior, and they quantitatively measured the clarity and complexity of the questions as well as the accuracy and appropriateness of the responses, applying a 3-tiered grading scale (low, medium, and high). Furthermore, a linguistic analysis was conducted to compare the length and vocabulary diversity of the responses using word count and type-token ratio. Results: GPT-4 and human experts displayed comparable efficacy in medical accuracy ("GPT-4 is better" at 132/251, 52.6% vs "Human expert is better" at 119/251, 47.4%). In accuracy level categorization, humans had more high-accuracy responses than GPT-4 (50/237, 21.1% vs 30/238, 12.6%) but also a greater proportion of low-accuracy responses (11/237, 4.6% vs 1/238, 0.4%; P=.001). GPT-4 responses were generally longer and used a less diverse vocabulary than those of human experts, potentially enhancing their comprehensibility for general users (sentence count: mean 10.9, SD 4.2 vs mean 5.9, SD 3.7; P<.001; type-token ratio: mean 0.69, SD 0.07 vs mean 0.79, SD 0.09; P<.001). Nevertheless, human experts outperformed GPT-4 in specific question categories, notably those related to drug or medication information and preliminary diagnoses. These findings highlight the limitations of GPT-4 in providing advice based on clinical experience. Conclusions: GPT-4 has shown promising potential in automated medical consultation, with comparable medical accuracy to human experts. However, challenges remain particularly in the realm of nuanced clinical judgment. Future improvements in LLMs may require the integration of specific clinical reasoning pathways and regulatory oversight for safe use. Further research is needed to understand the full potential of LLMs across various medical specialties and conditions.

Harnessing health information technology to promote equitable care for patients with limited English proficiency and complex care needs.

BACKGROUND: Patients with language barriers encounter healthcare disparities, which may be alleviated by leveraging interpreter skills to reduce cultural, language, and literacy barriers through improved bidirectional communication. Evidence supports the use of in-person interpreters, especially for interactions involving patients with complex care needs. Unfortunately, due to interpreter shortages and clinician underuse of interpreters, patients with language barriers frequently do not get the language services they need or are entitled to. Health information technologies (HIT), including artificial intelligence (AI), have the potential to streamline processes, prompt clinicians to utilize in-person interpreters, and support prioritization. METHODS: From May 1, 2023, to June 21, 2024, a single-center stepped wedge cluster randomized trial will be conducted within 35 units of Saint Marys Hospital & Methodist Hospital at Mayo Clinic in Rochester, Minnesota. The units include medical, surgical, trauma, and mixed ICUs and hospital floors that admit acute medical and surgical care patients as well as the emergency department (ED). The transitions between study phases will be initiated at 60-day intervals resulting in a 12-month study period. Units in the control group will receive standard care and rely on clinician initiative to request interpreter services. In the intervention group, the study team will generate a daily list of adult inpatients with language barriers, order the list based on their complexity scores (from highest to lowest), and share it with interpreter services, who will send a secure chat message to the bedside nurse. This engagement will be triggered by a predictive machine-learning algorithm based on a palliative care score, supplemented by other predictors of complexity including length of stay and level of care as well as procedures, events, and clinical notes. DISCUSSION: This pragmatic clinical trial approach will integrate a predictive machine-learning algorithm into a workflow process and evaluate the effectiveness of the intervention. We will compare the use of in-person interpreters and time to first interpreter use between the control and intervention groups. TRIAL REGISTRATION: NCT05860777. May 16, 2023.

Cognitive Predictors of the Overlap of Reading and Math in Middle School.

Math and reading skills are known to be related, and predictors of each are well researched. What is less understood is the extent to which these predictors, uniquely and collectively, overlap with one another, are differentially important for different academic skills, and account for the overlap of math and reading. We examined 20 potential predictors from four domains (working memory, processing speed, attention, and language) using latent variables and both timed and untimed achievement skill, in a sample (N=212) of at-risk middle schoolers, half of whom were English learners. The predictors accounted for about half of the overlap among achievement skills, suggesting that other factors (e.g., domain specific skills) might also be relevant for the overlap. We also found some differential prediction (language for reading, working memory for math). The present results extend and refine our understanding of the contribution of these cognitive predictors for reading and math.

Evaluation of large language models as a diagnostic aid for complex medical cases.

Background: The use of large language models (LLM) has recently gained popularity in diverse areas, including answering questions posted by patients as well as medical professionals. Objective: To evaluate the performance and limitations of LLMs in providing the correct diagnosis for a complex clinical case. Design: Seventy-five consecutive clinical cases were selected from the Massachusetts General Hospital Case Records, and differential diagnoses were generated by OpenAI's GPT3.5 and 4 models. Results: The mean number of diagnoses provided by the Massachusetts General Hospital case discussants was 16.77, by GPT3.5 30 and by GPT4 15.45 (p < 0.0001). GPT4 was more frequently able to list the correct diagnosis as first (22% versus 20% with GPT3.5, p = 0.86), provide the correct diagnosis among the top three generated diagnoses (42% versus 24%, p = 0.075). GPT4 was better at providing the correct diagnosis, when the different diagnoses were classified into groups according to the medical specialty and include the correct diagnosis at any point in the differential list (68% versus 48%, p = 0.0063). GPT4 provided a differential list that was more similar to the list provided by the case discussants than GPT3.5 (Jaccard Similarity Index 0.22 versus 0.12, p = 0.001). Inclusion of the correct diagnosis in the generated differential was correlated with PubMed articles matching the diagnosis (OR 1.40, 95% CI 1.25-1.56 for GPT3.5, OR 1.25, 95% CI 1.13-1.40 for GPT4), but not with disease incidence. Conclusions and relevance: The GPT4 model was able to generate a differential diagnosis list with the correct diagnosis in approximately two thirds of cases, but the most likely diagnosis was often incorrect for both models. In its current state, this tool can at most be used as an aid to expand on potential diagnostic considerations for a case, and future LLMs should be trained which account for the discrepancy between disease incidence and availability in the literature.

Experience in psychological counseling supported by artificial intelligence technology.

BACKGROUND: In recent years, artificial intelligence (AI) technology has been continuously advancing and finding extensive applications, with one of its core technologies, machine learning, being increasingly utilized in the field of healthcare. OBJECTIVE: This research aims to explore the role of Artificial Intelligence (AI) technology in psychological counseling and utilize machine learning algorithms to predict counseling outcomes. METHODS: Firstly, by employing natural language processing techniques to analyze user conversations with AI chatbots, researchers can gain insights into the psychological states and needs of users during the counseling process. This involves detailed analysis using text analysis, sentiment analysis, and other relevant techniques. Subsequently, machine learning algorithms are used to establish predictive models that forecast counseling outcomes and user satisfaction based on data such as user language, emotions, and behavior. These predictive results can assist counselors or AI chatbots in adjusting counseling strategies, thereby enhancing counseling effectiveness and user experience. Additionally, this study explores the potential and prospects of AI technology in the field of psychological counseling. RESULTS: The research findings indicate that the designed machine learning models achieve an accuracy rate of approximately 89% in analyzing psychological conditions. This demonstrates significant innovation and breakthroughs in AI technology. Consequently, AI technology will gradually become a highly important tool and method in the field of psychological counseling. CONCLUSION: In the future, AI chatbots will become more intelligent and personalized, providing users with precise, efficient, and convenient psychological counseling services. The results of this research provide valuable technical insights for further improving AI-supported psychological counseling, contributing positively to the application and development of AI technology.

Evaluation of Online AI-Generated Foot and Ankle Surgery Information.

As a natural progression from educational pamphlets to the worldwide web, and now artificial intelligence (AI), OpenAI chatbots provide a simple way of obtaining pathology-specific patient information, however, little is known concerning the readability and quality of foot and ankle surgery information. This investigation compares such information using the commercially available OpenAI ChatGPT Chatbot and FootCareMD®. A list of common foot and ankle pathologies from FootCareMD® were queried and compared with similar results using ChatGPT. From both resources, the Flesch Reading Ease Score (FRES) and Flesch-Kincaid Grade Level (FKGL) scores were calculated for each condition. Qualitative analysis of each query was performed using the JAMA Benchmark Criteria Score and the DISCERN Score.The overall ChatGPT and FootCareMD® FRES scores were 31.12±7.86 and 55.18±7.27, respectively (p<0.0001). The overall ChatGPT and FootCareMD® FKGL scores were 13.79±1.22 and 9.60±1.24 respectively (p<0.0001), except for the pilon fracture FKGL scores (p=0.09). The average JAMA Benchmark for all information obtained through ChatGPT and FootCareMD® were 0±0 and 1.95±0.15 (p < 0.001), respectively. The DISCERN Score for all information obtained through ChatGPT and FootCareMD® were 52.53±5.39 and 66.93±4.57 (p < 0.001), respectively. AI-assisted queries concerning common foot and ankle pathologies are written at a higher grade level and with less reliability and accuracy compared to similar information available on FootCareMD®. With the ease of use and increase in AI technology, consideration should be given to the nature and quality of information being shared with respect to the diagnosis and treatment of foot and ankle conditions. LEVEL OF EVIDENCE: IV.

An Overview of the ADReSS-M Signal Processing Grand Challenge on Multilingual Alzheimer's Dementia Recognition Through Spontaneous Speech.

The ADReSS-M Signal Processing Grand Challenge was held at the 2023 IEEE International Conference on Acoustics, Speech and Signal Processing, ICASSP 2023. The challenge targeted difficult automatic prediction problems of great societal and medical relevance, namely, the detection of Alzheimer's Dementia (AD) and the estimation of cognitive test scoress. Participants were invited to create models for the assessment of cognitive function based on spontaneous speech data. Most of these models employed signal processing and machine learning methods. The ADReSS-M challenge was designed to assess the extent to which predictive models built based on speech in one language generalise to another language. The language data compiled and made available for ADReSS-M comprised English, for model training, and Greek, for model testing and validation. To the best of our knowledge no previous shared research task investigated acoustic features of the speech signal or linguistic characteristics in the context of multilingual AD detection. This paper describes the context of the ADReSS-M challenge, its data sets, its predictive tasks, the evaluation methodology we employed, our baseline models and results, and the top five submissions. The paper concludes with a summary discussion of the ADReSS-M results, and our critical assessment of the future outlook in this field.

A bibliometric analysis of speech and language impairments in Parkinson's disease based on Web of Science.

Many individuals with Parkinson's disease suffer from speech and language impairments that significantly impact their quality of life. Despite several studies on these disorders, there is a lack of relevant bibliometric analyses. This paper conducted a bibliometric analysis of 3,610 papers on speech and language impairments in Parkinson's disease patients from January 1961 to November 2023, based on the Web of Science Core Collection database. Using Citespace software, the analysis focused on annual publication volume, cooperation among countries and institutions, author collaborations, journals, co-citation references, and keywords, aiming to explore the current research status, hotspots, and frontiers in this field. The number of annual publications related to speech and language impairment in Parkinson's disease have been increasing over the years. The USA leads in the number of publications. Research hotspots include the mechanism underlying speech and language impairments, clinical symptoms, automated diagnosis and classification of patients with PD using linguistic makers, and rehabilitation interventions.

Development of a culturally enhanced caregiver-facilitated language nutrition intervention "+Language is Medicine" to address developmental delay in Diné (Navajo) toddlers.

Introduction: Developmental Delay (DD) is highly common in American Indian and Alaska Native (AI/AN; Indigenous) toddlers and leads to high numbers of AI/AN children who eventually need special education services. AI/AN children are 2.89 times more likely to receive special education compared to other children in the U.S., yet developmental disorders are more frequently under diagnosed and untreated in AI/AN infants and toddlers. DD, which can be identified as early as toddlerhood, can lead to negative impacts on developmental trajectories, school readiness, and long-term health. Signs of DD can be identified early with proper developmental screening and remediated with high quality early intervention that includes effective parent training. There are many evidence-based language facilitation interventions often used in Early Intervention programs. However, in communities in rural parts of the Navajo Nation where there are limited services and resources, infants and toddlers with early signs of DD are often missed and do not get the culturally responsive support and evidence-based intervention they deserve. Methods: The community-based +Language is Medicine (+LiM) study team partnered with tribal home visitors, community members, and a Diné linguist/elder using a collaborative virtual workgroup approach in 2021 and 2022 to present the +LiM pilot study aims and to discuss strategies for enhancing a language intervention for toddlers experiencing DD in their tribal community. This paper will detail the stages of community engagement, intervention enhancement and preparation for field testing of the +LiM intervention to address elevated rates of DD in toddlers in the Northern Agency of the Navajo Nation. Results: Two major outcomes from this collaborative workgroup included: (1) a team-initiated redefining of language nutrition to align with Indigenous values that center cultural connectedness and native language use and (2) a five-lesson caregiver-facilitated curriculum titled +Language is Medicine which includes caregiver lessons on language nutrition, language facilitation, shared book reading, pretend play, and incorporation of native language into home routines. These two workgroup outcomes were leveraged to develop a pilot pre-/post-intervention study to test the effectiveness of the +LiM intervention with caregiver-toddler dyads living on the Navajo Nation. Discussion: Delivering tailored child interventions through tribal home visiting are cost-effective and innovative methods for reaching reservation-based families who benefit from culturally responsive parent coaching and instruction. The +LiM team has applied a precision tribal home visiting approach to enhance methods of early intervention for children with DD. Our enhancement process was grounded in Indigenous community-based participatory research that centered culture and language.

Effects of cue and instructor demonstration on the learning of Chinese characters for Chinese as a second language beginners.

This study purports to investigate the effects of cue and instructor demonstration on Chinese as a second language (CSL) beginners' Chinese character learning performance, cognitive load, learning motivation and attitude. In the current research, 100 CSL beginners were randomly assigned to four experimental groups, i.e., instructor demonstration cued character, instructor demonstration non-cued character, non-instructor demonstration cued character and non-instructor demonstration non-cued character. Participants were instructed to watch an instructional video and subsequently complete a post-test and a questionnaire. The results show that (1) in the presence of instructor demonstration, the cued characters can noticeably reduce CSL beginners' cognitive load and enhance their learning attitudes towards character learning, enabling them to achieve better performance in character stroke but not in radical and structure, and (2) in the presence of cued characters, the instructor demonstration can noticeably reduce CSL beginners' cognitive load and increase their learning motivation and attitudes towards character learning but can not improve their character learning performance. The findings have significant implications for educators and instructional designers of Chinese and other non-alphabetic languages, such as Kanji and Hangul.

Reshaping free-text radiology notes into structured reports with generative question answering transformers.

BACKGROUND: Radiology reports are typically written in a free-text format, making clinical information difficult to extract and use. Recently, the adoption of structured reporting (SR) has been recommended by various medical societies thanks to the advantages it offers, e.g. standardization, completeness, and information retrieval. We propose a pipeline to extract information from Italian free-text radiology reports that fits with the items of the reference SR registry proposed by a national society of interventional and medical radiology, focusing on CT staging of patients with lymphoma. METHODS: Our work aims to leverage the potential of Natural Language Processing and Transformer-based models to deal with automatic SR registry filling. With the availability of 174 Italian radiology reports, we investigate a rule-free generative Question Answering approach based on the Italian-specific version of T5: IT5. To address information content discrepancies, we focus on the six most frequently filled items in the annotations made on the reports: three categorical (multichoice), one free-text (free-text), and two continuous numerical (factual). In the preprocessing phase, we encode also information that is not supposed to be entered. Two strategies (batch-truncation and ex-post combination) are implemented to comply with the IT5 context length limitations. Performance is evaluated in terms of strict accuracy, f1, and format accuracy, and compared with the widely used GPT-3.5 Large Language Model. Unlike multichoice and factual, free-text answers do not have 1-to-1 correspondence with their reference annotations. For this reason, we collect human-expert feedback on the similarity between medical annotations and generated free-text answers, using a 5-point Likert scale questionnaire (evaluating the criteria of correctness and completeness). RESULTS: The combination of fine-tuning and batch splitting allows IT5 ex-post combination to achieve notable results in terms of information extraction of different types of structured data, performing on par with GPT-3.5. Human-based assessment scores of free-text answers show a high correlation with the AI performance metrics f1 (Spearman's correlation coefficients>0.5, p-values<0.001) for both IT5 ex-post combination and GPT-3.5. The latter is better at generating plausible human-like statements, even if it systematically provides answers even when they are not supposed to be given. CONCLUSIONS: In our experimental setting, a fine-tuned Transformer-based model with a modest number of parameters (i.e., IT5, 220 M) performs well as a clinical information extraction system for automatic SR registry filling task. It can extract information from more than one place in the report, elaborating it in a manner that complies with the response specifications provided by the SR registry (for multichoice and factual items), or that closely approximates the work of a human-expert (free-text items); with the ability to discern when an answer is supposed to be given or not to a user query.

Methods and approaches for developing plain language versions of guidelines recommendations: Scoping review protocol.

AIMS AND OBJECTIVES: This is a protocol of a scoping review that will aim to synthesise methodological evidence on formulating plain language versions of recommendations from guidelines both for clinical practice and for public health. METHOD: We will conduct a search in MEDLINE (Ovid), Embase (Ovid) databases, and webpages of guidelines developers with no language and date limitations. The title/abstract and full-text screening will be performed by two reviewers independently. The team of reviewers will extract data on methods used for developing plain language versions of recommendations in a standardised manner. The data analysis and synthesis will be presented narratively in tabular form. RESULTS AND CONCLUSION: We will conduct a scoping review based on this protocol.

Evaluation of a Pilot Spanish Language Program for Student Pharmacists.

OBJECTIVE: To present findings from an evaluation of the Spanish Language Track (SLT) for student pharmacists, which assessed student outcomes and feedback. METHODS: A mixed-methods program evaluation was conducted with the first cohort of the SLT members (N = 10). Participants completed pre/post-surveys and focus groups. Quantitative data analysis employed descriptive and frequency analysis, while qualitative data was thematically analyzed. RESULTS: With a focus on qualitative themes, quantitative results support themes one, two, and three based on findings from the self-assessment of participants' ability to speak and use the Spanish Language. Five themes were identified: (1) initial involvement and motivation to engage; (2) language skill development; (3) health-focused language immersion; (4) strong relationships within the SLT cohort; and (5) opportunities for improvement. CONCLUSION: Findings demonstrate students' active engagement with SLT while enhancing language skills through immersive experiences. Their connections with other cohort members, SLT team members, Colombian pharmacists, and bi-weekly patient appointment simulations were key contributors to learning outcomes while offering suggestions for programming. The SLT provides a foundational model for health professional programs to offer students opportunities to understand and practice language-concordant healthcare delivery to promote improved health outcomes in Spanish-speaking populations.

Assessing Laterality Errors in Radiology: Comparing Generative AI and Natural Language Processing.

PURPOSE: We compared the performance of generative AI (G-AI, ATARI) and natural language processing (NLP) tools for identifying laterality errors in radiology reports and images. METHODS: We used an NLP-based (mPower) tool to identify radiology reports flagged for laterality errors in its QA Dashboard. The NLP model detects and highlights laterality mismatches in radiology reports. From an initial pool of 1124 radiology reports flagged by the NLP for laterality errors, we selected and evaluated 898 reports that encompassed radiography, CT, MRI, and ultrasound modalities to ensure comprehensive coverage. A radiologist reviewed each radiology report to assess if the flagged laterality errors were present (reporting error - true positive) or absent (NLP error - false positive). Next, we applied ATARI to 237 radiology reports and images with consecutive NLP true positive (118 reports) and false positive (119 reports) laterality errors. We estimated accuracy of NLP and G-AI tools to identify overall and modality-wise laterality errors. RESULTS: Among the 898 NLP-flagged laterality errors, 64% (574/898) had NLP errors and 36% (324/898) were reporting errors. The text query ATARI feature correctly identified the absence of laterality mismatch (NLP false positives) with a 97.4% accuracy (115/118 reports; 95% CI = 96.5% - 98.3%). Combined Vision and text query resulted in 98.3% accuracy (116/118 reports/images; 95% CI = 97.6% - 99.0%) query alone had a 98.3% accuracy (116/118 images; 95% CI = 97.6% - 99.0%). CONCLUSION: The generative AI-empowered ATARI prototype outperformed the assessed NLP tool for determining true and false laterality errors in radiology reports while enabling an image-based laterality determination. Underlying errors in ATARI text query in complex radiology reports emphasize the need for further improvement in the technology.

Reasoning with large language models for medical question answering.

OBJECTIVES: To investigate approaches of reasoning with large language models (LLMs) and to propose a new prompting approach, ensemble reasoning, to improve medical question answering performance with refined reasoning and reduced inconsistency. MATERIALS AND METHODS: We used multiple choice questions from the USMLE Sample Exam question files on 2 closed-source commercial and 1 open-source clinical LLM to evaluate our proposed approach ensemble reasoning. RESULTS: On GPT-3.5 turbo and Med42-70B, our proposed ensemble reasoning approach outperformed zero-shot chain-of-thought with self-consistency on Steps 1, 2, and 3 questions (+3.44%, +4.00%, and +2.54%) and (2.3%, 5.00%, and 4.15%), respectively. With GPT-4 turbo, there were mixed results with ensemble reasoning again outperforming zero-shot chain-of-thought with self-consistency on Step 1 questions (+1.15%). In all cases, the results demonstrated improved consistency of responses with our approach. A qualitative analysis of the reasoning from the model demonstrated that the ensemble reasoning approach produces correct and helpful reasoning. CONCLUSION: The proposed iterative ensemble reasoning has the potential to improve the performance of LLMs in medical question answering tasks, particularly with the less powerful LLMs like GPT-3.5 turbo and Med42-70B, which may suggest that this is a promising approach for LLMs with lower capabilities. Additionally, the findings show that our approach helps to refine the reasoning generated by the LLM and thereby improve consistency even with the more powerful GPT-4 turbo. We also identify the potential and need for human-artificial intelligence teaming to improve the reasoning beyond the limits of the model.