Tag: nevin manimala

Epilepsy Behav. 2025 Nov 23;174:110805. doi: 10.1016/j.yebeh.2025.110805. Online ahead of print.

ABSTRACT

BACKGROUND: The efficacy of Managing Epilepsy Well Network (MEWN) self-management programs is well-established. The purpose of this study was to replicate the HOBSCOTCH program to evaluate program implementation and assess effectiveness on patient cognition, quality of life and self-management behaviors.

METHODS: Participants from clinical and community settings were recruited and randomized to intervention vs waitlist control. Program outcomes assessing quality of life, cognition, treatment adherence, depressive symptoms, and self-management behavior were measured at baseline and 3 months. Program staff provided survey data guided by the RE-AIM model regarding program delivery and acceptability, appropriateness, and feasibility. Data were analyzed using descriptive statistics, linear regression, and qualitative methods.

RESULTS: This predominantly female (69 %) and White (64 %) sample (N = 61) with active epilepsy (66 % had a seizure in the past year) also had a higher portion of Black (32 %) participants. Analyses yielded significant differences in cognition, quality of life and self-management behaviors between the two groups at follow-up. Staff indicated the packaged intervention, remote delivery, trained coaches and investment of leaders/clinical staff as intervention benefits. Implementation barriers included difficulty contacting participants and mental health concerns. Program appropriateness and feasibility ratings were high; 98 % reported that they very much or moderately enjoyed working with their coach and would recommend the program to others.

CONCLUSION: HOBSCOTCH was effective in increasing cognition and quality of life in people with epilepsy in this replication study. A novel finding highlighted changes in the frequency of participants’ self-management behaviors. These findings have implications for healthcare systems incorporating evidence-based self-management programs for their patients.

PMID:41285072 | DOI:10.1016/j.yebeh.2025.110805

Int J Med Inform. 2025 Nov 20;207:106190. doi: 10.1016/j.ijmedinf.2025.106190. Online ahead of print.

ABSTRACT

BACKGROUND: Despite the increasing use of AI tools like ChatGPT, Claude, and Gemini in scientific writing, concerns remain about their ability to generate accurate, high-quality, and consistent abstracts for research publications. The reliability of AI-generated abstracts in dental research is questionable when compared to human-written counterparts. This study aimed to develop a framework for evaluating AI-generated abstracts and compare the performance of ChatGPT, Claude, and Gemini against human-written abstracts in dental research.

METHODS: The DAISY framework was developed to evaluate AI-generated abstracts across five domains: Data accuracy (D), Abstract quality (A), Integrity and consistency (I), Syntax and fluency (S), and Yield of human likelihood (Y). Reliability of the framework was assessed using Cohens Kappa (κ = 0.85) and Pearsons’s correlation coefficient (0.92) for inter- and intra- expert reliability and was found to be satisfactory. This study adopted a comparative observational study design. Eight research articles belonging to structured (n = 4) and unstructured (n = 4) categories were selected from reputable journals. Researchers trained in scientific writing wrote abstracts for these articles, while AI-generated abstracts were obtained using specific prompts. Ten dental experts evaluated the abstracts using this framework. Statistical analysis was performed using ANOVA and Tukey’s post-hoc test.

RESULTS: Human-written abstracts consistently outperformed AI-generated ones across all DAISY framework domains. Among AI tools, ChatGPT scored highest in all DAISY framework domains, followed by Gemini and Claude. Human-written abstracts achieved the highest human likelihood score (90.25 ± 4.68), while AI-generated abstracts scored below 50%, with Gemini scoring least (3.25 ± 1.75). The differences between the groups were statistically significant (P ≤ 0.05).

CONCLUSION: The DAISY framework proved reliable for evaluating AI-generated abstracts. While ChatGPT performed better than other AI tools, none matched the quality of human-written abstracts. This indicates that AI tools, though valuable, remain limited in producing credible scientific writing in dental research.

PMID:41285065 | DOI:10.1016/j.ijmedinf.2025.106190

Immunohorizons. 2025 Nov 24;9(12):vlaf064. doi: 10.1093/immhor/vlaf064.

ABSTRACT

Ficolins, encoded by FCN genes, are key pattern recognition molecules of the lectin complement pathway involved in immune complex clearance, a process often impaired in systemic lupus erythematosus (SLE). Genetic polymorphisms in FCN genes may influence disease susceptibility. However, their functional significance in SLE remains unclear. The present study aimed to investigate the association of selected FCN gene single-nucleotide polymorphisms (SNPs) with SLE, lupus nephritis (LN), and serum ficolin levels in a Western Indian cohort. Seven SNPs in FCN1 (rs2989727, rs1071583), FCN2 (rs7851696, rs17549193, rs7865453, rs17514136), and FCN3 (rs3813800) were genotyped in 200 SLE patients and 200 healthy controls using polymerase chain reaction (PCR) sequence-specific primer and PCR restriction fragment length polymorphism. Serum ficolin-1, -2, and -3 levels were measured using ELISA. Statistical analysis included χ2 test, Kruskal-Wallis test, and logistic regression to assess associations and calculate odds ratios with 95% confidence intervals. The analysis identified significant associations of FCN2 rs7851696, rs7865453, and rs17514136, as well as FCN3 rs3813800, with SLE susceptibility. Among LN patients, FCN1 rs2989727 and rs1071583, FCN2 rs17514136, and FCN3 rs3813800 showed significant associations. FCN3 rs3813800 was significantly associated with ficolin-3 levels, while FCN2 rs7865453 was associated with complement component 1q-circulation immune complex levels. These findings provide novel insight into associations of FCN gene polymorphisms with SLE and LN susceptibility, with genotype-phenotype correlations suggesting their biological relevance. Future longitudinal and mechanistic studies are warranted to validate these associations and explore their therapeutic potential.

PMID:41285030 | DOI:10.1093/immhor/vlaf064

Hum Reprod. 2025 Nov 24:deaf218. doi: 10.1093/humrep/deaf218. Online ahead of print.

ABSTRACT

STUDY QUESTION: Is male infertility independently associated with an increased risk of incident hypertension, ischemic and non-ischemic heart disease, diabetes, and/or cerebrovascular disease?

SUMMARY ANSWER: Fathers diagnosed with male infertility have a modestly increased risk of heart disease, diabetes, and hypertension compared with fertile fathers, after controlling for measured confounders; however, some important confounders remain inadequately measured.

WHAT IS KNOWN ALREADY: Cohort studies suggest that infertile men have an increased risk of incident cardiometabolic diseases, including diabetes, hypertension, heart disease, and cerebrovascular disease, although findings are mixed. The reasons for this association are unclear, but cardiometabolic conditions and male infertility share a wide range of shared etiological factors including age, chronic conditions such as obesity and obstructive sleep apnea, cancers and their treatments, environmental exposures such as pollution and pesticides, lifestyle factors such as smoking and cardiorespiratory fitness, autoimmune conditions such as lupus and Hashimoto’s thyroiditis, as well as congenital conditions such as cystic fibrosis and muscular dystrophy.

STUDY DESIGN, SIZE, DURATION: Our population-based cohort study included 445 909 men whose partner conceived a child between January 2009 and September 2016 in New South Wales (NSW), Australia. We excluded men with a diagnosis of infertility prior to 2009, men who were under the age of 14 at the time of the child’s conception, and men diagnosed with cardiometabolic conditions in the 6.5 years prior to their index date. The index date was the later of the date of the child’s conception or the date of the vasectomy for fertile men or the date of diagnosis of infertility for infertile men, i.e. the time when the exposure status was determined. From the index date, we followed participants for 5 years up until the latest available date of September 2021.

PARTICIPANTS/MATERIALS, SETTINGS, METHODS: The study was conducted in NSW, Australia. We determined infertility status by a diagnosis of male infertility in the Australian and New Zealand Assisted Reproduction Database, hospital records, or a record of fertility-related procedures. We assessed the following outcomes: incident hypertension, ischemic and non-ischemic heart disease, all heart disease, diabetes, and cerebrovascular disease. We calculated age-standardized prevalence rates at baseline. We mapped potential confounding pathways using directed acyclic graphs and controlled for measured confounders using inverse probability of treatment weighting and g-computation. We estimated adjusted marginal risk ratios (aRR) and adjusted marginal risk differences (aRD) using robust Poisson regression.

MAIN RESULTS AND THE ROLE OF CHANCE: The number of events and 5-year crude incidence rate for the outcomes were: hypertension (events: 17 433, fertile: 41.09 per 1000 population, infertile: 70.03 per 1000 population), all heart disease (events: 15 549, fertile: 36.44 per 1000 population, infertile: 59.88 per 1000 population), ischemic heart disease (events: 12 628 fertile: 29.24 per 1000 population, infertile: 47.1 per 1000 population), non-ischemic heart disease (events: 5183, fertile: 11.69 per 1000 population, infertile: 20.24 per 1000 population), cerebrovascular disease (events: 512, fertile: 1.14 per 1000 population, infertile: 1.78 per 1000 population) and diabetes (events: 7064, fertile: 16.05 per 1000 population, infertile: 27.59 per 1000 population). Compared with fertile men, men diagnosed with infertility demonstrated increased risk of incident disease for: hypertension aRR = 1.20 (95% CI 1.11-1.31, P < 0.001), aRD = 1.1% (95% CI: 0.6%-1.6%, P < 0.001); all heart disease aRR = 1.20 (95% CI 1.09-1.31, P < 0.001), aRD =0.9% (95% CI: 0.4%-1.4%, P < 0.001); non-ischemic heart disease aRR = 1.26 (95% CI 1.08-1.48, P = 0.004), aRD = 0.4% (95% CI: 0.1%-0.7%, P = 0.009); ischemic heart disease aRR = 1.13 (95% CI 1.02-1.25, P = 0.020), aRD = 0.4% (95% CI: 0.1%-0.7%, P = 0.028); and diabetes aRR = 1.28 (95% CI 1.12-1.46, P < 0.001), aRD 0.6% (0.2%-0.9%, P = 0.001). There was no significant difference in the incidence of cerebrovascular disease, aRR = 1.0 (95% CI 0.56-1.80, P = 0.996), aRD = 0.0% (95% CI: -0.1% to 0.1%, P = 0.996). These results remained consistent in sensitivity analyses, including an expanded exposure definition of infertility, a 10-year follow-up period, changing the outcomes of people who died in follow-up, and using an alternative index date.

LIMITATIONS, REASONS FOR CAUTION: The cohort includes men who fathered a child, so men who did not seek to, or were unable to, have a child, and men with poor access to the reproductive healthcare may not be included. This may generate selection effects, biasing the estimates toward the null. We were unable to adequately control for several confounders, including important lifestyle factors like smoking, diet, cardiorespiratory fitness, and alcohol intake, due to data limitations, which may bias estimates away from the null. It appears plausible that a combination of unmeasured and inadequately measured confounders may attenuate the observed estimates.

WIDER IMPLICATIONS OF THE FINDINGS: These findings suggest that male infertility may serve as an early indicator for a slightly heightened cardiometabolic risk, specifically relating to hypertension, diabetes, and various forms of heart disease. Our study is the largest on this topic, with extensive control for confounders. Our findings align with published research, indicating that men diagnosed with infertility have a slightly higher risk of incident diabetes, hypertension, and heart disease. From a public health perspective, fertility treatment may be an opportunity for earlier detection and intervention to help prevent the onset of cardiometabolic conditions in men diagnosed with infertility, particularly given that men generally have low rates of contact with the health system.

STUDY FUNDING/COMPETING INTEREST(S): The PhD candidacy of J.M. is supported by Medical Research Future Fund (MRFF) Emerging Priorities and Consumer Driven Research initiative: EPCD000007, 2020. M.K.O’B. and G.M.C. declare receiving payment to their institution by the same MRFF grant. G.M.C. reports receiving funding from an Australian MRFF grant paid to UNSW to support this work, and J.M. reports receiving PhD funding from the same MRFF grant. C.V. declares an unpaid role on Human Reproduction’s Editorial Board, and paid employment at the University of New South Wales (UNSW) until January 2023. The National Perinatal Epidemiology and Statistics Unit (NPESU), which belongs to UNSW, is custodian of the Australian and New Zealand Assisted Reproduction Database (ANZARD). Data from ANZARD were used in this study. G.M.C. also declares paid employment from UNSW. The remaining authors have nothing to declare.

TRIAL REGISTRATION NUMBER: N/A.

PMID:41285026 | DOI:10.1093/humrep/deaf218

JMIR Form Res. 2025 Nov 24;9:e80090. doi: 10.2196/80090.

ABSTRACT

BACKGROUND: Physical inactivity increases the risk of chronic disease and reduces life expectancy, yet adherence to physical activity (PA) guidelines remains low. SMS text messages are promising for promoting PA, but it is not clear what type of messaging is most effective. Messages with causal information, which explain why a recommendation is being made, may be more persuasive than messages containing only recommendations.

OBJECTIVE: This study aims to compare the effectiveness of causal versus noncausal SMS text messages for promoting PA in US adults.

METHODS: In this pilot study, we randomized US adults (n=28 in the analytic sample) aged 18-64 years to receive causal or noncausal SMS text messages roughly every other day for 2 weeks, following a 1-week baseline. PA was measured using Empatica wristbands during intervention and baseline periods, and the International Physical Activity Questionnaire – Short Form (IPAQ-SF) at baseline, postintervention, and 4 weeks later. The primary outcome was the change in mean metabolic equivalent of tasks (METs) per minute from baseline to intervention. The secondary outcomes were (1) PA differences on intervention and nonintervention days (mean METs/min), (2) changes in self-reported METs per week between surveyed periods, and (3) participant satisfaction. We used a linear mixed model to analyze our primary outcome, the Mann-Whitney U test and the chi-square test of independence to analyze quantitative secondary outcomes, and qualitative coding to analyze survey data.

RESULTS: The causal message group had a greater increase in mean METs per minute from baseline to intervention compared to the noncausal group with a moderate effect size (P=.01; Cohen d=0.54). In the causal group, PA was significantly higher on SMS text message days (mean 2.46, SD 0.12 METs/min) compared to nonmessage days (mean 2.25, SD 0.15 METs/min; P=.02), while there was no difference in the noncausal group (P=.54). No significant between-group difference was found in self-reported PA or satisfaction.

CONCLUSIONS: Causal information that links suggested PA to health outcomes can increase the effectiveness of SMS text messages promoting PA, indicating the value of incorporating causal information into intervention design. Our results provide further basis for just-in-time interventions, as activity was higher on message days. Further work is needed to better personalize message content and timing to maintain participant engagement.

PMID:41284987 | DOI:10.2196/80090

J Med Internet Res. 2025 Nov 24;27:e77155. doi: 10.2196/77155.

ABSTRACT

BACKGROUND: Osteoporosis is a prevalent skeletal disorder characterized by decreased bone mass and increased fracture risk; however, it frequently remains underdiagnosed due to limited health care resources and its asymptomatic progression. Deep learning (DL) provides a promising solution for automated screening using computed tomography (CT) scans, enabling earlier detection and improved management.

OBJECTIVE: This systematic review and meta-analysis aimed to investigate the diagnostic performance of DL models in diagnosing osteoporosis based on CT scans.

METHODS: This study was conducted under the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines using articles extracted from PubMed, Scopus, Web of Science (Core), and Embase (Ovid). Studies involving adult participants who underwent CT and in which DL was applied for osteoporosis diagnosis were included. The QUADAS-2 (Quality Assessment of Diagnostic Accuracy Studies-2) tool was used to estimate the risk of bias in each study. The confusion matrices from the included studies were extracted to summarize the diagnostic performance of DL models for osteoporosis. Within a bivariate random-effects framework, sensitivity and specificity were jointly synthesized to yield the summary estimates. Heterogeneity was quantified with Higgins I² statistics. Subgroup analyses were performed to explore potential sources of heterogeneity among the included studies.

RESULTS: This review included 24 studies, encompassing CT images from 29,808 participants. All studies used conventional CT scans and used DL-based architectures. Fifteen, 6, and 3 studies were assessed as having a low, uncertain, and high risk of bias, respectively. The meta-analysis included 20 studies. The pooled sensitivity and specificity were 0.88 (95% CI 0.85-0.91; I2=83.69%) and 0.94 (95% CI 0.91-0.96; I2=95.07%) for osteoporosis diagnosis; 0.81 (95% CI 0.76-0.85; I2=82.38%) and 0.92 (95% CI 0.90-0.94; I2=79.05%) for osteopenia identification; and 0.95 (95% CI 0.92-0.97; I2=98.28%) and 0.93 (95% CI 0.91-0.95; I2=94.93%) for normal case identification. The area under the curve of the DL models for identifying osteoporosis, osteopenia, and normal cases was 0.96 (95% CI 0.93-0.97), 0.94 (95% CI 0.92-0.96), and 0.98 (95% CI 0.96-0.99), respectively. Subgroup analyses revealed that models based on DenseNet variants (P<.01), multislice input (P<.01), 3D architecture (P<.01), and CT as the reference standard (P<.01) demonstrated superior diagnostic performance.

CONCLUSIONS: This study indicated that CT-based DL models achieve promising diagnostic performance for osteoporosis. However, substantial heterogeneity among the included studies, limited external validation, and incomplete end-to-end pipelines constrain the generalizability of the proposed models. Further research is warranted to support their clinical translation and standardized application.

PMID:41284986 | DOI:10.2196/77155

JMIR Form Res. 2025 Nov 24;9:e66931. doi: 10.2196/66931.

ABSTRACT

BACKGROUND: Breast cancer remains the most common cancer among women globally. Mammography is a key diagnostic modality; however, interpretation is increasingly challenged by rising imaging volumes, a global shortage of breast radiologists, and variability in reader experience. Artificial intelligence (AI) has been proposed as a potential adjunct to address these issues, particularly in settings with high breast density, such as Asian populations. This study aimed to evaluate the impact of AI assistance on mammographic diagnostic performance among resident and consultant radiologists in Singapore.

OBJECTIVE: To assess whether AI assistance improves diagnostic accuracy in mammographic breast cancer detection across radiologists with varying levels of experience.

METHODS: A multi-reader, multi-case study was conducted at the National University Hospital, Singapore, from May to August 2023. De-identified digital mammograms from 500 women (250 with cancer and 250 normal or benign) were interpreted by 17 radiologists (4 consultants, 4 senior residents, and 9 junior residents). Each radiologist read all cases over 2 reading sessions: one without AI assistance and another with AI assistance, separated by a 1-month washout period. The AI system (FxMammo) provided heatmaps and malignancy risk scores to support decision-making. Area under the curve of the receiver operating characteristic (AUROC) was used to evaluate diagnostic performance.

RESULTS: Among the 500 cases, 250 were malignant and 250 were non-malignant. Of the malignant cases, 16%(80/500) were ductal carcinoma in situ and 84%(420/500) were invasive cancers. Among non-malignant cases, 69.2%(346/500) were normal, 17.6%(88) benign, and 3.6%(18/500) possibly benign but stable on follow-up. Masses (54.4%, 272) and calcifications (10.8%, 54/500) were the most common findings in cancer cases. A majority of both malignant (66.8%, 334/500) and non-malignant (68%, 340/500) cases had heterogeneously or extremely dense breasts (BIRADS [Breast Imaging Reporting and Data System] categories C and D). The AI model achieved an AUROC of 0.93 (95% CI 0.91-0.95), slightly higher than consultant radiologists (AUROC 0.90, 95% CI 0.89-0.92; P=.21). With AI assistance, AUROC improved among junior residents (from 0.84 to 0.86; P=.38) and senior residents (from 0.85 to 0.88; P=.13), with senior residents approaching consultant-level performance (AUROC difference 0.02; P=.051). Diagnostic gains with AI were greatest in women with dense breasts and among less experienced radiologists. AI also improved inter-reader agreement and time efficiency, particularly in benign or normal cases.

CONCLUSIONS: This is the first study in Asia to evaluate AI assistance in mammography interpretation by radiologists of varying experience. AI significantly improved diagnostic performance and efficiency among residents, helping to narrow the experience-performance gap without compromising specificity. These findings suggest a role for AI in enhancing diagnostic consistency, improving workflow, and supporting training. Integration into clinical and educational settings may offer scalable benefits, though careful attention to threshold calibration, feedback loops, and real-world validation remains essential. Further studies in routine screening settings are needed to confirm generalizability and cost-effectiveness.

PMID:41284978 | DOI:10.2196/66931

JMIR Mhealth Uhealth. 2025 Nov 24;13:e73903. doi: 10.2196/73903.

ABSTRACT

BACKGROUND: Mobile health (mHealth) apps are believed to be an effective method to support family caregivers to better care for patients with stroke. This study’s purpose was to explore the status and the influencing factors of mHealth app use among family caregivers of patients with stroke via machine learning (ML) models.

OBJECTIVE: This study aimed to understand the status quo of mHealth app use among community family caregivers of patients with stroke and the factors influencing their use behavior. Six ML models were used to construct the classifier, and the Shapley Additive Explanations (SHAP) algorithm was introduced to interpret the best ML model.

METHODS: In this cross-sectional study, family carers of patients with stroke were recruited. Data on their basic profile and mHealth app use were obtained through face-to-face questionnaires. Hedonic motivation, usage habits, and other relevant information were additionally measured among app users. A total of 12 models were constructed using six ML algorithms. The top-performing logistic regression and random forest models were further analyzed with SHAP to interpret key influencing factors.

RESULTS: A total of 360 family caregivers of patients with stroke were included in this study from March 2023 to November 2023, of which 206 (57.2%) reported having used mHealth apps. Of the 6 ML models, the logistic regression model performed the best in terms of whether caregivers used the mHealth app, with an area under the receiver operating characteristic curve of 0.753 (95% CI 0.698-0.802), accuracy of 0.694 (95% CI 0.647-0.742), sensitivity of 0.748 (95% CI 0.688-0.806), and specificity of 0.623 (95% CI 0.547-0.698). SHAP analysis showed that the top 5 most influencing factors were educational level, age, the patient’s self-care ability, the relationship with the cared-for individual, and the duration of illness. The random forest model performed best in terms of use behavior with an area under the receiver operating characteristic curve of 0.773 (95% CI 0.725-0.818), accuracy of 0.602 (95% CI 0.534-0.665), sensitivity of 0.476 (95% CI 0.420-0.533), and specificity of 0.769 (95% CI 0.738-0.797). The SHAP analysis revealed that hedonic motivation, habits, occupation, convenience conditions, and effort expectations were the 5 most significant influencing factors.

CONCLUSIONS: The research results indicate that the software developers and policymakers of mHealth apps should take the abovementioned influencing factors into consideration when developing and promoting the software. We should focus on the older adults with lower educational levels, lower the threshold for software use, and provide more convenient conditions. By grasping the hedonistic tendencies and habitual usage characteristics of users, they can provide them with more concise and accurate health information, which will enhance the popularity and effectiveness of mHealth apps.

PMID:41284965 | DOI:10.2196/73903

Hepatol Commun. 2025 Nov 24;9(12):e0856. doi: 10.1097/HC9.0000000000000856. eCollection 2025 Dec 1.

ABSTRACT

BACKGROUND: Dietary interventions are key for managing metabolic dysfunction-associated steatotic liver disease (MASLD), yet optimal diets and the role of PNPLA3 in modulating response to diet remain unclear. We evaluated the efficacy of a Mediterranean diet (MD) versus a low-fat diet (LFD) on hepatic fat and fibrosis, assessing interactions with PNPLA3 genotype.

METHODS: Two hundred fifty adults with MASLD with BMI ≥25 kg/m2 were randomized to a 12-week moderately hypocaloric MD or LFD intervention. Individuals with excess alcohol intake and other etiologies of steatosis were excluded. Subjects were genotyped for PNLPA3 single-nucleotide polymorphism. Anthropometric measures, blood tests, and liver assessments [controlled attenuation parameter (CAP) and liver stiffness measurement (LSM)] were conducted at baseline and follow-up. Essential food items were provided, and adherence was tracked using validated questionnaires. The primary outcome was CAP, analyzed using linear mixed models adjusted for age and metabolic syndrome.

RESULTS: Both diets significantly reduced CAP, LSM, and body weight at follow-up, with no significant differences between groups. The mean difference between MD and LFD was -0.13 dB/m for CAP (p=0.976, 95% CI: -8.54, 8.28), -0.19 kPa for LSM (p=0.355, 95% CI: -0.58, 0.21), and 3.01 kg for weight (p=0.159, 95% CI: -7.21, 1.19). PNPLA3 genotype did not significantly interact with diet for CAP, LSM, or weight (p=0.286, p=0.464, p=0.622, respectively).

CONCLUSIONS: Weight reduction achieved by MD and LFD is similarly efficient in steatosis and fibrosis reduction, while PNPLA3 genotype does not affect the response to diet. Further studies investigating the impact of diet and nutrigenetics on liver-related outcomes are warranted.

PMID:41284948 | DOI:10.1097/HC9.0000000000000856

Integr Environ Assess Manag. 2025 Nov 24:vjaf175. doi: 10.1093/inteam/vjaf175. Online ahead of print.

ABSTRACT

This study investigates the environmental and economic factors driving pollution-related mortality in 37 Sub-Saharan African countries from 1990 to 2022. The analysis combines two complementary approaches. The first is a Panel Panel Quantile Autoregressive Distributed Lag (QARDL) model, which captures both short- and long-run relationships across different levels of the mortality distribution. The second is a Light Gradient Boosting Machine (LightGBM) model, a machine-learning method that detects nonlinear patterns and reveals interactions that may be missed by traditional statistical models. Together, these methods integrate structured econometric inference with flexible pattern recognition, offering a clearer and more reliable picture of how environmental and economic forces jointly shape mortality outcomes. The LightGBM partial dependence plots further confirm the Panel QARDL results, showing consistent directional effects across all variables. Fine particulate matter and consumer price index display the strongest nonlinear responses, while methane, health expenditures and Gross Domestic Product exhibit moderate but coherent patterns that reinforce the robustness of the findings. The results show that higher levels of fine particulate matter are consistently linked to increased mortality across all quantiles. Economic growth reduces mortality at higher quantiles of the distribution, where health burdens are most severe, indicating that stronger economies are better able to mitigate pollution-related deaths. Inflation exhibits a positive relationship with mortality, particularly at higher quantiles, indicating that rising prices can limit access to essential health services and increase vulnerability. Health spending, in contrast, reduces mortality, though its impact varies by both time horizon and income level. Overall, the findings highlight the importance of cleaner air, stable prices and stronger healthcare systems for reducing pollution-related mortality in Sub-Saharan Africa. The study provides policy-relevant insights for promoting health resilience under economic and environmental stress.

PMID:41284938 | DOI:10.1093/inteam/vjaf175