Nevin Manimala

Eye Contact Lens. 2025 Aug 15. doi: 10.1097/ICL.0000000000001216. Online ahead of print.

ABSTRACT

OBJECTIVES: This study aimed to estimate direct and indirect economic costs associated with Acanthamoeba keratitis and explore variations in individual and healthcare characteristics.

METHODS: Patients treated at Moorfields Eye Hospital, UK, between January 2011 and August 2014 were surveyed. Patients were included if they had a confirmed diagnosis and wore contact lenses. Direct costs included healthcare expenses, whereas indirect costs encompassed lost wages for patients and caregivers. The Mann-Whitney U test and generalized linear models were used to analyze cost data and their associated factors, with P<0.05 being considered statistically significant.

RESULTS: A total of 73 patients were included, with a mean age of 39.3 years (19-81); 40 patients (54.8%) were female, and 41 (56.2%) had good outcomes. Median direct, indirect, and total costs were £2,966.5 (IQR £5,553.2), £1,776.0 (IQR £3,223.3), and £6,678.6 (IQR £17,268.4), respectively. Higher direct costs were independently associated with multiple medical visits (P<0.001), steroids before antiamoeba therapy (P=0.042), and keratoplasty (P<0.001).

CONCLUSIONS: The cost of managing Acanthamoeba keratitis is substantial. Future research should focus on optimizing treatment strategies and improving patient outcomes to help alleviate the financial burden on patients and healthcare systems.

PMID:40829159 | DOI:10.1097/ICL.0000000000001216

JMIR Form Res. 2025 Aug 19;9:e73193. doi: 10.2196/73193.

ABSTRACT

BACKGROUND: Social media is an important source of e-cigarette-related information for adolescents and young adults. However, misinformation is being shared across platforms, which may encourage e-cigarette use.

OBJECTIVE: This study aims to examine adolescent and young adult perceptions of a novel pop-up on social media that provides links to vaping-related health information from credible sources (eg, CDC).

METHODS: Between December 2023 and March 2024, participants aged 13-24 years (N=5326) completed a web-based cross-sectional survey. Participants were asked to select from 4 positive (eg, useful to share, helpful to check health-related information) and 4 negative or neutral sentiments (eg, usually ignore such notifications, would not trust it) to reflect their perceptions about a mock pop-up that provided a link to e-cigarette-related information on social media.

RESULTS: More participants endorsed positive sentiments than negative or neutral sentiments in relation to the pop-up. Specifically, 1078 (20.8%) endorsed two or more positive sentiments, and 690 (13.3%) endorsed two or more negative or neutral sentiments when searching for “vaping” on social media; similarly 913 (17.6%) participants endorsed two or more positive sentiments and 690 (13.3%) endorsed two or more negative or neutral sentiments when viewing e-cigarette-related posts in their feed (all P<.001). Among those who were searching for e-cigarette-related information, participants aged 13-18 years were more likely to endorse at least two positive sentiments compared to those aged 19-24 years (ie, 646, 22.0% vs 423, 19.2%, respectively), those who had never used e-cigarettes compared with those who had ever used them (ie, 674, 23.6% vs 404, 17.3%, respectively), and those who last used e-cigarettes more than 30 days ago compared with those who had used them in the past 30 days (ie, 187, 19.8% vs 217, 15.6%, respectively). Similarly, among participants who viewed e-cigarette-related posts in their feed, those who had never used e-cigarettes were more likely to endorse two or more positive perceptions compared to those who had ever used e-cigarettes (ie, 563, 19.7% vs 350, 15.0%), and those who had last used e-cigarettes more than 30 days ago (ie, 159, 16.9% vs 191, 13.7%) were more likely to endorse two or more positive perceptions compared to those who had used them in the past 30 days (all Ps<.001). Participants who had never used e-cigarettes were also less likely to trust pop-ups, compared to those who had ever used them, whether while searching for e-cigarette-related information on social media or while seeing e-cigarette-related posts in their feed (approximately, 19%vs 24%). There were no significant differences in the endorsement of negative or neutral sentiments. However, participants aged 13-18 years were more likely to ignore such notifications while viewing e-cigarette-related posts in their social media feed compared to those aged 19-24 years (ie, 850, 28.9% vs 563 25%); further, participants who had never used e-cigarettes were less trusting of e-cigarette-related information while searching for vaping or in their feed.

CONCLUSIONS: Positive perception of a social media pop-up indicates its potential to prevent e-cigarette-related misinformation. Further development of a pop-up requires strategies to better engage and inform adolescents and young adults, specifically younger individuals, as they may be more likely to disregard pop-ups, and older individuals and those who used e-cigarettes in the past 30 days, as they are potentially more dismissive of such information.

PMID:40829154 | DOI:10.2196/73193

JMIR Form Res. 2025 Aug 19;9:e70863. doi: 10.2196/70863.

ABSTRACT

BACKGROUND: Thyroid nodules are common, with ultrasound imaging as the primary modality for their assessment. Risk stratification systems like the American College of Radiology Thyroid Imaging Reporting and Data System (ACR TI-RADS) have been developed but suffer from interobserver variability and low specificity. Artificial intelligence, particularly large language models (LLMs) with multimodal capabilities, presents opportunities for efficient end-to-end diagnostic processes. However, their clinical utility remains uncertain.

OBJECTIVE: This study evaluates the accuracy and consistency of multimodal LLMs for thyroid nodule risk stratification using the ACR TI-RADS system, examining the effects of model fine-tuning, image annotation, prompt engineering, and comparing open-source versus commercial models.

METHODS: In total, 3 multimodal vision-language models were evaluated: Microsoft’s open-source Large Language and Visual Assistant (LLaVA) model, its medically fine-tuned variant (Large Language and Vision Assistant for bioMedicine [LLaVA-Med]), and OpenAI’s commercial o3 model. A total of 192 thyroid nodules from publicly available ultrasound image datasets were assessed. Each model was evaluated using 2 prompts (basic and modified) and 2 image scenarios (unlabeled vs radiologist-annotated), yielding 6912 responses. Model outputs were compared with expert ratings for accuracy and consistency. Statistical comparisons included Chi-square tests, Mann-Whitney U tests, and Fleiss’ kappa for interrater reliability.

RESULTS: Overall, 88.4% (6110/6912) of responses were valid, with the o3 model producing the highest validity rate (2273/2304, 98.6%), followed by LLaVA (2108/2304, 91.5%) and LLaVA-Med (1729/2304, 75%; P<.001). The o3 model demonstrated the highest accuracy overall, achieving up to 57.3% accuracy in Thyroid Imaging Reporting and Data System (TI-RADS) classification, although still remaining suboptimal. Labeled images improved accuracy marginally in nodule margin assessment only when evaluating LLaVA models (407/768, 53% to 447/768, 58.2%; P=.04). Prompt engineering improved accuracy for composition (649/1,152, 56.3% vs 483/1152, 41.9%; P<.001), but significantly reduced accuracy for shape, margins, and overall classification. Consistency was the highest with the o3 model (up to 85.4%), but was comparable for LLaVA and significantly improved with image labeling and modified prompts across multiple TI-RADS categories (P<.001). Subgroup analysis for o3 alone showed prompt engineering did not affect accuracy significantly but markedly improved consistency across all TI-RADS categories (up to 97.1% for shape, P<.001). Interrater reliability was consistently poor across all combinations (Fleiss’ kappa<0.60).

CONCLUSIONS: The study demonstrates the comparative advantages and limitations of multimodal LLMs for thyroid nodule risk stratification. While the commercial model (o3) consistently outperformed open-source models in accuracy and consistency, even the best-performing model outputs remained suboptimal for direct clinical deployment. Prompt engineering significantly enhanced output consistency, particularly in the commercial model. These findings underline the importance of strategic model optimization techniques and highlight areas requiring further development before multimodal LLMs can be reliably used in clinical thyroid imaging workflows.

PMID:40829145 | DOI:10.2196/70863

J Med Internet Res. 2025 Aug 19;27:e69031. doi: 10.2196/69031.

ABSTRACT

BACKGROUND: University students frequently face mental health challenges due to academic pressures, lifestyle changes, and developmental factors. Digital interventions, such as Doing What Matters in Times of Stress (DWM), a psychosocial e-mental health intervention developed by the World Health Organization (WHO), offer scalable approaches to address these issues. These data emerging from the literature provide the framework for the CAMPUS (Characterize and Address Mental health Problems in University Students) study aimed at supporting the mental health of students attending the University of Verona.

OBJECTIVE: This study aimed to assess the effectiveness and implementability of DWM as a psychological strategy for effective mental health prevention and promotion, as well as for reducing psychological symptoms and distress and improving well-being in university students.

METHODS: During the study period (October 2023-June 2024), we conducted a prospective hybrid type-1 nonrandomized follow-up study, with a pretest-posttest design. The study population consisted of students attending the University of Verona, who were recruited through university communication channels and participated via web-based platforms. Data were collected at baseline (T1) and after the intervention (T2) using an ad hoc sociodemographic information page and self-reported tools assessing psychological distress with the Kessler-10 (K-10), depressive symptoms with the Patient Health Questionnaire-9 (PHQ-9) depression scale, anxiety symptoms with the Generalized Anxiety Disorder-7 (GAD-7) scale, and psychological well-being with the WHO-5 Well-Being Index (WHO-5). In addition, at postintervention, the implementability was assessed. Statistical analyses included Wilcoxon matched pairs signed rank tests and logistic regression models to identify associated factors.

RESULTS: Out of 2296 interested students, 1498 (65.24%) completed all DWM sessions and assessments. At T1, students exhibited mild psychological distress, anxiety, and depressive symptoms with moderate well-being. Significant improvements were observed postintervention: the K-10 scores decreased from 22.41 (SD 6.54) to 19.86 (SD 5.96), the GAD-7 scale scores decreased from 8.27 (SD 4.31) to 6.57 (SD 3.76), and the PHQ-9 scores decreased from 8.28 (SD 7.73) to 6.75 (SD 4.37; all P<.001). The WHO-5 well-being scores increased from 11.73 (SD 4.65) to 13.26 (SD 4.68; P<.001). Satisfaction was high, with 90.72% (1359/1498) of participants agreeing or strongly agreeing on satisfaction, 77.37% (1159/1498) agreeing or strongly agreeing on appropriateness, and 94.99% (1423/1498) finding the program easy to use. No significant differences in clinical outcomes were associated with sociodemographic or baseline mental health variables.

CONCLUSIONS: The DWM intervention demonstrated positive effects on students’ mental health, showing reductions in distress, anxiety, and depressive symptoms, alongside improved well-being. The program’s high levels of acceptability, appropriateness, and feasibility highlight its potential for broader application as a digital mental health strategy for university students.

PMID:40829124 | DOI:10.2196/69031

JMIR Form Res. 2025 Aug 19;9:e73078. doi: 10.2196/73078.

ABSTRACT

BACKGROUND: Disaster telehealth can be used to provide rapid access to remote specialty expertise and virtual surge capacity for overwhelmed local clinicians. The Regional Disaster Health Response System (RDHRS) is developing a disaster teleconsultation system for cross-jurisdictional care in the United States. In 2020, the Region 1 RDHRS provided Massachusetts hospitals access to disaster teleconsultation services with out-of-state critical care experts during the first wave of the COVID-19 pandemic response.

OBJECTIVE: We aimed to field-test (1) the acceptability and feasibility of using a prototype, web-based disaster teleconsultation platform with minimal-to-no user training and (2) the feasibility of deploying a national volunteer expert pool to access out-of-state expertise.

METHODS: This was a prospective, mixed methods, observational study. We recruited field clinicians from Massachusetts hospitals and out-of-state critical-care physicians as experts for a 2-week pilot (June 2020). Experts were trained to use a prototype platform, while field clinicians received a just-in-time tool. Field clinicians requested teleconsultations for hospitalized patients with COVID-19 (clinical call) or simulated patients (test call). We collected demographics, call performance data, and Telehealth Usability Questionnaire (TUQ) ratings to measure acceptability (primary outcome; total usability score ≥6 of 7) and feasibility (secondary outcome; interface, interaction quality, and reliability items), and interviewed participants. We report descriptive statistics and key themes using the Technology Acceptance Model framework.

RESULTS: Ten experts from 6 states and 17 field clinicians from 4 hospitals participated. All experts and 10 field clinicians completed postpilot questionnaires (74% response overall). Of these, 20% had previously used telemedicine in a disaster. In total, 50 test calls and no clinical calls were logged. Most (70%) made ≥1 call; 22% (95% CI 10%-34%) connected successfully. The median time to connect was 1.6 (IQR 3.2) minutes. Among field clinician respondents, 50% used smartphone devices, 40% hospital desktop computers, and 10% laptop computers to access RDHRS teleconsultation services. Calls failed due to platform routing errors (49%), hospital computers without cameras or microphones (10%), firewalls (8%), and expert notification failures (5%). The mean total usability score was 5.6 (SD 1.3). TUQ item scores were highest in usefulness (mean 6.0, SD 1.1) and ease-of-use (mean 6.0, SD 1.4), and lowest in reliability (mean 2.4, SD 1.4). Participants were comfortable using the platform. Those with difficulty identified discomfort with technology as the cause. All experts were willing to participate in a national expert registry and obtain emergency licensure, and most (80%) were willing to serve on a volunteer, unpaid basis.

CONCLUSIONS: Clinicians found the prototype platform acceptable, but the workflow requires revision to reduce call failure and improve feasibility and reliability for future use with minimal-to-no training. Using familiar clinical workflows for emergency consultation and mobile devices with camera and microphone capabilities could improve call performance and reliability.

PMID:40829123 | DOI:10.2196/73078

Neurology. 2025 Sep 9;105(5):e213996. doi: 10.1212/WNL.0000000000213996. Epub 2025 Aug 19.

ABSTRACT

BACKGROUND AND OBJECTIVES: While preclinical studies documented potential neuroprotective effects of estrogen/progesterone, the impact of biological sex on outcomes after traumatic spinal cord injury (tSCI) remains incompletely understood. The aim of this study was to compare male and female age-stratified subgroups, which presumably correspond to premenopausal, perimenopausal, and postmenopausal states, concerning access to optimal care and their outcomes after tSCI.

METHODS: A series of propensity score-matched cohort studies was performed, comparing female individuals with male individuals in the younger (age≤ 40 years), middle-aged (41-50 years), and older (>50 years) subgroups regarding baseline data; management strategies; access to optimal care; and hospital, neurologic, and functional outcomes after tSCI. Data were selected from 5,571 individuals with tSCI at C1-L2 who were enrolled in the Rick Hansen Spinal Cord Injury Registry from July 2004 to September 2019. Outcome measures included in-hospital mortality and motor and sensory score changes from admission to an acute care hospital to discharge from a rehabilitation center.

RESULTS: In the younger subgroups, female individuals (n = 320, mean age: 26 years) were more often White and had a greater proportion of tSCIs due to falls or transportation-related accidents than male individuals (n = 320, mean age: 26.5 years). Younger female and male subgroups had similar in-hospital mortality rates (0.6% vs 0.6%, p = 1) and motor (4 vs 5, p = 0.8919) and sensory (2 vs 5, p = 0.5) score changes after tSCI. In the middle-aged subgroups, female individuals (n = 133, mean age: 46 years) and male individuals (n = 113, mean age: 46 years) had statistically comparable baseline data, in-hospital mortality rates (2.3% vs 3.0%, p = 1), and motor (12 vs 5, p = 0.8766) and sensory (0 vs 2, p = 0.4918) score changes. In the older subgroups, female individuals (n = 531, mean age: 68 years) showed a higher frequency of fall-related tSCI but they had similar in-hospital mortality rates (7.7% vs 9.8%, p = 0.2324) and motor (9 vs 8, p = 0.7594) and sensory (5 vs 0, p = 0.7204) score changes after tSCI compared with male individuals (n = 531, mean age: 68 years).

DISCUSSION: The results of this study suggest that biological sex does not significantly affect in-hospital mortality and neurologic recovery after tSCI when optimal care is applied, according to data from a Canadian registry. Those findings support the notion that sex equity in management strategies promotes equal outcomes after tSCI.

PMID:40829102 | DOI:10.1212/WNL.0000000000213996

Neurology. 2025 Sep 9;105(5):e213975. doi: 10.1212/WNL.0000000000213975. Epub 2025 Aug 19.

ABSTRACT

BACKGROUND AND OBJECTIVES: Higher white matter hyperintensity volume (WMHV) is associated with hippocampal atrophy, cognitive decline, and dementia; however, it is unknown whether continually increasing WMHV is related to hippocampal atrophy. The aim of this study was to determine whether higher WMHV change rate (WMHVR) is related to higher hippocampal atrophy rate (HAR); this relationship is dependent on cardiovascular risk factors (CVRFs), Alzheimer disease (AD) pathology, and genetic risk; and this relationship is mediated by neuroaxonal degradation.

METHODS: Participants from Insight46, a substudy of the 1946 British Birth Cohort, underwent combined [¹⁸F]florbetapir PET/MRI scans at University College London approximately 2.5 years apart. WMHVR was quantified from T2/fluid-attenuated inversion recovery and HAR from T1 sequences. Life-course blood pressure and CVRF data were measured at 6 and 3 time points, respectively. APOE genotype and neurofilament light chain (NfL) quantification were derived from blood samples. Participants with neurologic conditions were excluded from primary analyses. Linear regression was used to test the relationships between WMHVR and HAR, adjusting for sex, age, and total intracranial volume (TIV) and CVRF, APOE-ε4 status, and β-amyloid (Aβ) in separate models. Semipartial R² was calculated from these models. In a post hoc analysis, structural equation modeling aimed to determine whether NfL mediated the relationship between WMHVR and HAR.

RESULTS: A total of 317 individuals without neurologic conditions (48% female, 100% White British, mean baseline age [SD] = 70.5 [0.6] years) were included. The mean HAR was 0.04 [0.04] mL/y. Each 1 mL/y increase in WMHVR was associated with a 0.014 mL/y (95% CI 0.005-0.022) increase in HAR, adjusted for TIV, age, and sex (p = 0.002). Adjustment for additional variables did not meaningfully attenuate this association (≥0.012 mL/y, p ≤ 0.005, all models), and there was no indirect effect through NfL (0.0004 mL/y [95% CI -0.0006 to 0.0012], p < 0.1).

DISCUSSION: Higher WMHVR was associated with HAR between approximately 70 and 73 years, independent of CVRF levels, APOE-ε4 status, and Aβ load, and not mediated by markers of neuroaxonal degradation. Although AD-specific pathology is typically considered the main cause of accelerated HAR, we demonstrated that HAR is also linked to deteriorating WM health. These results will need to be replicated in more diverse cohorts with longer follow-up periods to confirm the findings.

PMID:40829101 | DOI:10.1212/WNL.0000000000213975

Am J Manag Care. 2025 Aug 1;31(8):e238-e240. doi: 10.37765/ajmc.2025.89780.

ABSTRACT

OBJECTIVE: To assess trends in assigned International Statistical Classification of Diseases, Tenth Revision ( ICD-10 ) codes for patients hospitalized with heart failure (HF) from 2018 to 2022 in 2 large US health systems.

STUDY DESIGN: Retrospective cross-sectional analysis of ICD-10 codes assigned to patients hospitalized with HF in the Providence Health and University of Colorado Health (UCHealth) systems.

METHODS: The study included patients discharged from the Providence Health and UCHealth systems between 2018 and 2022 with a primary diagnosis of HF. ICD-10 codes analyzed included systolic HF (I50.2), diastolic HF (I50.3), combined systolic and diastolic HF (I50.4), hypertensive heart disease with HF (I11.0), and hypertensive heart disease with HF and chronic kidney disease (CKD) (I13.0, I13.2). Hospitalization data were analyzed separately for each health system due to privacy policies.

RESULTS: Between 2018 and 2022, 61,238 HF hospitalizations occurred in the Providence Health system, and 13,576 occurred in UCHealth. Hypertensive heart disease with HF and CKD was the most common diagnosis, accounting for 42% to 56% of HF hospitalizations, followed by hypertensive heart disease with HF (34%-42%). Together, these diagnoses represented 85% to 90% of HF hospitalizations. Systolic, diastolic, and combined HF codes accounted for only 9% to 18% of hospitalizations. Significant variability in hypertension prevalence (ie, 100% in Providence Health and 38%-39% in UCHealth) was observed between the 2 health systems in patients with codes I13.0 and I13.2.

CONCLUSIONS: The study highlighted a significant shift in HF diagnosis codes, with hypertensive heart disease with HF with and without CKD now predominant. The findings highlight the need for standardized coding practices across health systems for quality improvement initiatives and health services research.

PMID:40829099 | DOI:10.37765/ajmc.2025.89780

Mil Med. 2025 Aug 19:usaf407. doi: 10.1093/milmed/usaf407. Online ahead of print.

ABSTRACT

INTRODUCTION: The Defense Health Agency directs that military Medical Treatment Facilities provide Patient-Centered Medical Home (PCMH) services within its Internal Medicine and other Primary Care clinics. Development and documentation of self-management goals for chronic diseases within the electronic medical record (EMR) is a PCMH accreditation standard through the Joint Commission. This quality improvement project aimed to implement a structured approach to creating and documenting SMART (Specific, Measurable, Achievable, Relevant, and Time-bound) goals for patients with Type 2 Diabetes Mellitus (DM) in a military-affiliated Internal Medicine clinic. Although SMART goals have demonstrated clinically meaningful reductions in hemoglobin A1c (HbA1c), the effects of SMART goals on patient perceptions of their health and self-management of their DM have not been fully studied.

MATERIALS AND METHODS: This study voluntarily enrolled 62 patients aged 19-90 with DM and HbA1c >8% (64 mmol/mol) within the previous 6-month period from a military-affiliated Internal Medicine clinic. Exclusion criteria included patients following with a subspecialist for DM management. Enrolled patients developed and discussed SMART goals with their physician using a standardized goal-setting tool for self-management of their DM. These goals were documented in the EMR for longitudinal follow-up between patients, physicians, and the clinic care team. HbA1c and patient self-reported health perceptions were evaluated before and after 3 months of participation in this intervention. Data analysis used a 2-tailed paired samples t-test. The study protocol underwent Institutional Review Board examination and was determined not to constitute research.

RESULTS: Of 62 patients enrolled, 36 completed post-surveys and were included in the statistical analysis. Results showed a statistically significant reduction in HbA1c levels, with a mean decrease from 9.18% (77.2 mmol/mol) to 7.84% (62.7 mmol/mol) (P < 0.001). Patient-reported perceptions of their health, goal-oriented behaviors, perceived ability to influence health through lifestyle changes, and responsibility for self-management of their DM did not change significantly.

CONCLUSIONS: This quality improvement project further highlights the importance of self-management goals in improving biomedical outcomes and disease management markers. Despite a significant improvement in HbA1c, no changes were observed in patient-reported health perceptions. Overall, patients indicated high adherence and positive feedback with this intervention. The goal-setting tool and standardized documentation strategy as implemented are a viable strategy for satisfying a PCMH accreditation standard. Future studies could address limitations such as single-site design, response rates, and review specific effects on perceptions with the use of validated survey instruments. Similar self-management interventions can be applied to other chronic diseases where daily lifestyle choices influence outcomes.

PMID:40829048 | DOI:10.1093/milmed/usaf407

Chaos. 2025 Aug 1;35(8):083127. doi: 10.1063/5.0279141.

ABSTRACT

We develop a rigorous mathematical framework to determine the optimal inter-layer edge configurations that maximize synchronizability in two-layer chain networks-an area previously limited to empirical approaches. Departing from prior work relying on numerical simulations, we analytically prove that synchronizability is maximized when inter-layer edges are placed (i) at the chain’s midpoint (single-edge case) and (ii) at the one-quarter and three-quarter positions (dual-edge case). We also compute the coupling strength thresholds and further conjecture the optimal placement pattern for an arbitrary number of inter-layer edges, supporting this hypothesis with extensive numerical validation. These results bridge spectral graph theory and network topology design, offering principled guidelines for engineering interconnected chain-like systems, such as supply chains, information dissemination, and epidemic spread.

PMID:40829032 | DOI:10.1063/5.0279141