RESUMO
AIM: Machine learning may be a tool with the potential for obesity prediction. This study aims to review the literature on the performance of machine learning models in predicting obesity and to quantify the pooled results through a meta-analysis. DATA SYNTHESIS: A systematic review and meta-analysis were conducted, including studies that used machine learning to predict obesity. Searches were conducted in October 2023 across databases including LILACS, Web of Science, Scopus, Embase, and CINAHL. We included studies that utilized classification models and reported results in the Area Under the ROC Curve (AUC) (PROSPERO registration: CRD42022306940), without imposing restrictions on the year of publication. The risk of bias was assessed using an adapted version of the Transparent Reporting of a multivariable prediction model for individual Prognosis or Diagnosis (TRIPOD). Meta-analysis was conducted using MedCalc software. A total of 14 studies were included, with the majority demonstrating satisfactory performance for obesity prediction, with AUCs exceeding 0.70. The random forest algorithm emerged as the top performer in obesity prediction, achieving an AUC of 0.86 (95%CI: 0.76-0.96; I2: 99.8%), closely followed by logistic regression with an AUC of 0.85 (95%CI: 0.75-0.95; I2: 99.6%). The least effective model was gradient boosting, with an AUC of 0.77 (95%CI: 0.71-0.82; I2: 98.1%). CONCLUSION: Machine learning models demonstrated satisfactory predictive performance for obesity. However, future research should utilize more comparable data, larger databases, and a broader range of machine learning models.
Assuntos
Aprendizado de Máquina , Obesidade , Valor Preditivo dos Testes , Humanos , Obesidade/diagnóstico , Obesidade/epidemiologia , Masculino , Feminino , Idoso , Fatores de Risco , Medição de Risco , Pessoa de Meia-Idade , Adulto , Fatores Etários , Reprodutibilidade dos Testes , Técnicas de Apoio para a Decisão , Adulto Jovem , Idoso de 80 Anos ou mais , Diagnóstico por Computador , PrognósticoRESUMO
BACKGROUND: The UK Biobank is a large middle-aged cohort recruited in 2006-2010. We used data from its participants to analyze mortality, survival, and causes of death associated with mental disorders. METHODS: Our exposures were mental disorders identified using (1) symptom-based outcomes derived from an online Mental Health Questionnaire (n = 157 329), including lifetime/current depression, lifetime/current generalized anxiety disorder, lifetime/recent psychotic experience, lifetime bipolar disorder, current alcohol use disorder, and current posttraumatic stress disorder and (2) hospital data linkage of diagnoses within the International Classification of Diseases, 10th revision (ICD-10) (n = 502 422), including (A) selected diagnoses or groups of diagnoses corresponding to symptom-based outcomes and (B) all psychiatric diagnoses, grouped by ICD-10 section. For all exposures, we estimated age-adjusted mortality rates and hazard ratios, as well as proportions of deaths by cause. RESULTS: We found significantly increased mortality risk associated with all mental disorders identified by symptom-based outcomes, except for lifetime generalized anxiety disorder (with hazard ratios in the range of 1.08-3.0). We also found significantly increased mortality risk associated with all conditions identified by hospital data linkage, including selected ICD-10 diagnoses or groups of diagnoses (2.15-7.87) and ICD-10 diagnoses grouped by section (2.02-5.44). Causes of death associated with mental disorders were heterogeneous and mostly natural. CONCLUSIONS: In a middle-aged cohort, we found a higher mortality risk associated with most mental disorders identified by symptom-based outcomes and with all disorders or groups of disorders identified by hospital data linkage of ICD-10 diagnoses. The majority of deaths associated with mental disorders were natural.
Assuntos
Transtornos Mentais , Transtornos de Estresse Pós-Traumáticos , Pessoa de Meia-Idade , Humanos , Estudos Prospectivos , Causas de Morte , Bancos de Espécimes Biológicos , Transtornos Mentais/diagnóstico , Reino Unido/epidemiologiaRESUMO
PURPOSE OF REVIEW: To provide an overview of the literature regarding the use of machine learning algorithms to predict hypertension. A systematic review was performed to select recent articles on the subject. RECENT FINDINGS: The screening of the articles was conducted using a machine learning algorithm (ASReview). A total of 21 articles published between January 2018 and May 2021 were identified and compared according to variable selection, train-test split, data balancing, outcome definition, final algorithm, and performance metrics. Overall, the articles achieved an area under the ROC curve (AUROC) between 0.766 and 1.00. The algorithms most frequently identified as having the best performance were support vector machines (SVM), extreme gradient boosting (XGBoost), and random forest. Machine learning algorithms are a promising tool to improve preventive clinical decisions and targeted public health policies for hypertension. However, technical factors such as outcome definition, availability of the final code, predictive performance, explainability, and data leakage need to be consistently and critically evaluated.
Assuntos
Hipertensão , Algoritmos , Área Sob a Curva , Humanos , Hipertensão/diagnóstico , Aprendizado de Máquina , Máquina de Vetores de SuporteRESUMO
BACKGROUND: Populational ageing has been increasing in a remarkable rate in developing countries. In this scenario, preventive strategies could help to decrease the burden of higher demands for healthcare services. Machine learning algorithms have been increasingly applied for identifying priority candidates for preventive actions, presenting a better predictive performance than traditional parsimonious models. METHODS: Data were collected from the Health, Well Being and Aging (SABE) Study, a representative sample of older residents of São Paulo, Brazil. Machine learning algorithms were applied to predict death by diseases of respiratory system (DRS), diseases of circulatory system (DCS), neoplasms and other specific causes within 5 years, using socioeconomic, demographic and health features. The algorithms were trained in a random sample of 70% of subjects, and then tested in the other 30% unseen data. RESULTS: The outcome with highest predictive performance was death by DRS (AUC-ROC = 0.89), followed by the other specific causes (AUC-ROC = 0.87), DCS (AUC-ROC = 0.67) and neoplasms (AUC-ROC = 0.52). Among only the 25% of individuals with the highest predicted risk of mortality from DRS were included 100% of the actual cases. The machine learning algorithms with the highest predictive performance were light gradient boosted machine and extreme gradient boosting. CONCLUSION: The algorithms had a high predictive performance for DRS, but lower for DCS and neoplasms. Mortality prediction with machine learning can improve clinical decisions especially regarding targeted preventive measures for older individuals.
Assuntos
Doenças Cardiovasculares , Aprendizado de Máquina , Idoso , Algoritmos , Brasil/epidemiologia , Causas de Morte , HumanosRESUMO
BACKGROUND: Recent decreases in neonatal mortality have been slower than expected for most countries. This study aims to predict the risk of neonatal mortality using only data routinely available from birth records in the largest city of the Americas. METHODS: A probabilistic linkage of every birth record occurring in the municipality of São Paulo, Brazil, between 2012 e 2017 was performed with the death records from 2012 to 2018 (1,202,843 births and 447,687 deaths), and a total of 7282 neonatal deaths were identified (a neonatal mortality rate of 6.46 per 1000 live births). Births from 2012 and 2016 (N = 941,308; or 83.44% of the total) were used to train five different machine learning algorithms, while births occurring in 2017 (N = 186,854; or 16.56% of the total) were used to test their predictive performance on new unseen data. RESULTS: The best performance was obtained by the extreme gradient boosting trees (XGBoost) algorithm, with a very high AUC of 0.97 and F1-score of 0.55. The 5% births with the highest predicted risk of neonatal death included more than 90% of the actual neonatal deaths. On the other hand, there were no deaths among the 5% births with the lowest predicted risk. There were no significant differences in predictive performance for vulnerable subgroups. The use of a smaller number of variables (WHO's five minimum perinatal indicators) decreased overall performance but the results still remained high (AUC of 0.91). With the addition of only three more variables, we achieved the same predictive performance (AUC of 0.97) as using all the 23 variables originally available from the Brazilian birth records. CONCLUSION: Machine learning algorithms were able to identify with very high predictive performance the neonatal mortality risk of newborns using only routinely collected data.
Assuntos
Mortalidade Infantil , Morte Perinatal , Declaração de Nascimento , Brasil/epidemiologia , Feminino , Humanos , Recém-Nascido , Aprendizado de Máquina , GravidezRESUMO
PURPOSE: Our understanding of community-level predictors of individual mental disorders in large urban areas of lower income countries is limited. In particular, the proportion of migrant, unemployed, and poorly educated residents in neighborhoods of these urban areas may characterize group contexts and shape residents' health. METHODS: Cross-sectional household interviews of 7251 adults were completed across 83 neighborhoods of Buenos Aires, Argentina; Medellín, Colombia; São Paulo, Brazil; Lima, Peru; and Mexico City, Mexico as part of the World Mental Health Survey Initiative. Past-year internalizing and externalizing mental disorders were assessed, and multilevel models were used. RESULTS: Living in neighborhoods with either an above-average or below-average proportion of migrants and highly educated residents was associated with lower odds of any internalizing disorder (for proportion migrants: OR 0.75, 95% CI 0.62-0.91 for the bottom tertile and OR 0.79, 95% CI 0.67-0.94 for the top tertile compared to the middle tertile; for proportion highly educated: OR 0.76, 95% CI 0.64-0.90 for the bottom tertile and OR 0.58, 95% CI 0.37-0.90 for the top tertile compared to the middle tertile). Living in neighborhoods with an above-average proportion of unemployed individuals was associated with higher odds of having any internalizing disorder (OR 1.49, 95% CI 1.14-1.95 for the top tertile compared to the middle tertile). The proportion of highly educated residents was associated with lower odds of externalizing disorder (OR 0.54, 95% CI 0.31-0.93 for the top tertile compared to the middle tertile). CONCLUSIONS: The associations of neighborhood-level migration, unemployment, and education with individual-level odds of mental disorders highlight the importance of community context for understanding the burden of mental disorders among residents of rapidly urbanizing global settings.
Assuntos
Transtornos Mentais/epidemiologia , Pobreza/psicologia , Características de Residência/estatística & dados numéricos , Fatores Socioeconômicos , População Urbana/estatística & dados numéricos , Adulto , Argentina/epidemiologia , Brasil/epidemiologia , Cidades/epidemiologia , Colômbia/epidemiologia , Estudos Transversais , Escolaridade , Feminino , Inquéritos Epidemiológicos , Humanos , América Latina/epidemiologia , Masculino , Transtornos Mentais/psicologia , México/epidemiologia , Pessoa de Meia-Idade , Análise Multinível , Peru/epidemiologia , Migrantes/psicologia , Desemprego/psicologia , UrbanizaçãoRESUMO
BACKGROUND: Identifying successful public health ideas and practices is a difficult challenge towing to the presence of complex baseline characteristics that can affect health outcomes. We propose the use of machine learning algorithms to predict life expectancy at birth, and then compare health-related characteristics of the under- and overachievers (i.e., municipalities that have a worse and better outcome than predicted, respectively). METHODS: Our outcome was life expectancy at birth for Brazilian municipalities, and we used as predictors 60 local characteristics that are not directly controlled by public health officials (e.g., socioeconomic factors). RESULTS: The highest predictive performance was achieved by an ensemble of machine learning algorithms (cross-validated mean squared error of 0.168), including a 35% gain in comparison with standard decision trees. Overachievers presented better results regarding primary health care, such as higher coverage of the massive multidisciplinary program Family Health Strategy. On the other hand, underachievers performed more cesarean deliveries and mammographies and had more life-support health equipment. CONCLUSIONS: The findings suggest that analyzing the predicted value of a health outcome may bring insights about good public health practices.
Assuntos
Aprendizado de Máquina , Saúde Pública/estatística & dados numéricos , Idoso , Algoritmos , Brasil/epidemiologia , Cidades/epidemiologia , Humanos , Expectativa de Vida , Prática de Saúde Pública/estatística & dados numéricosRESUMO
OBJECTIVES: To estimate birth reduction potentially in response to Zika virus-associated microcephaly among the 36 largest Brazilian cities. METHODS: We analyzed the number of live births per month on the basis of information on approximately 8.2 million births from all of Brazil's state capitals and cities that had more than 10 000 annual births. RESULTS: In the second half of 2016, the live birth rate was reduced by 7.78% (95% confidence interval [CI] = 6.64%, 8.89%; P < .001). This reduction was correlated with the Zika virus-associated microcephaly rate. In the cities with the highest microcephaly rate in 2015 (> 1 case per 1000 live births), the reduction in the live birth rate was 10.84% (95% CI = 8.58%, 13.04%). CONCLUSIONS: The birth rate in the largest Brazilian cities during the second half of 2016 was significantly reduced, which is potentially the effect of a birth control recommendation prompted by an epidemiological alert. Public Health Implications. The effects of population-based interventions should be weighed by considering the actual risk of disease and the sociodemographic impact of strategies such as birth control.
Assuntos
Coeficiente de Natalidade , Epidemias/estatística & dados numéricos , Microcefalia/epidemiologia , Infecção por Zika virus/epidemiologia , Brasil/epidemiologia , Cidades/epidemiologia , Cidades/estatística & dados numéricos , Anticoncepção/estatística & dados numéricos , Feminino , Humanos , Microcefalia/etiologia , Microcefalia/virologia , Gravidez , Complicações Infecciosas na Gravidez/epidemiologia , Complicações Infecciosas na Gravidez/virologia , População Urbana/estatística & dados numéricos , Zika virus , Infecção por Zika virus/complicaçõesRESUMO
BACKGROUND: Clinicopathological studies are important in determining the brain lesions underlying dementia. Although almost 60% of individuals with dementia live in developing countries, few clinicopathological studies focus on these individuals. We investigated the frequency of neurodegenerative and vascular-related neuropathological lesions in 1,092 Brazilian admixed older adults, their correlation with cognitive and neuropsychiatric symptoms, and the accuracy of dementia subtype diagnosis. METHODS AND FINDINGS: In this cross-sectional study, we describe clinical and neuropathological variables related to cognitive impairment in 1,092 participants (mean age = 74 y, 49% male, 69% white, and mean education = 4 y). Cognitive function was investigated using the Clinical Dementia Rating (CDR) and the Informant Questionnaire on Cognitive Decline in the Elderly (IQCODE); neuropsychiatric symptoms were evaluated using the Neuropsychiatric Inventory (NPI). Associations between neuropathological lesions and cognitive impairment were investigated using ordinal logistic regression. We developed a neuropathological comorbidity (NPC) score and compared it to CDR, IQCODE, and NPI scores. We also described and compared the frequency of neuropathological diagnosis to clinical diagnosis of dementia subtype. Forty-four percent of the sample met criteria for neuropathological diagnosis. Among these participants, 50% had neuropathological diagnoses of Alzheimer disease (AD), and 35% of vascular dementia (VaD). Neurofibrillary tangles (NFTs), hippocampal sclerosis, lacunar infarcts, hyaline atherosclerosis, siderocalcinosis, and Lewy body disease were independently associated with cognitive impairment. Higher NPC scores were associated with worse scores in the CDR sum of boxes (ß = 1.33, 95% CI 1.20-1.46), IQCODE (ß = 0.14, 95% CI 0.13-0.16), and NPI (ß = 1.74, 95% CI = 1.33-2.16). Compared to neuropathological diagnoses, clinical diagnosis had high sensitivity to AD and high specificity to dementia with Lewy body/Parkinson dementia. The major limitation of our study is the lack of clinical follow-up of participants during life. CONCLUSIONS: NFT deposition, vascular lesions, and high NPC scorewere associated with cognitive impairment in a unique Brazilian sample with low education. Our results confirm the high prevalence of neuropathological diagnosis in older adults and the mismatch between clinical and neuropathological diagnoses.
Assuntos
Demência/epidemiologia , Idoso , Idoso de 80 Anos ou mais , Doença de Alzheimer/epidemiologia , Doença de Alzheimer/patologia , Brasil/epidemiologia , Cognição , Estudos Transversais , Demência/patologia , Demência Vascular/epidemiologia , Demência Vascular/patologia , Feminino , Humanos , Masculino , Pessoa de Meia-IdadeRESUMO
BACKGROUND: Brazil has one of the highest adolescent fertility rates in the world. Income inequality has been frequently linked to overall adolescent health, but studies that analyzed its association with adolescent fertility have been performed only in developed countries. Brazil, in the past decade, has presented a rare combination of increasing per capita income and decreasing income inequality, which could influence future desirable pathways for other countries. METHODS: We analyzed every live birth from 2000 and from 2010 in each of the 5,565 municipalities of Brazil, a total of 6,049,864 births, which included 1,247,145 (20.6%) births from women aged 15 to 19. Income inequality was assessed by the Gini Coefficient and adolescent fertility by the ratio between the number of live births from women aged 15 to 19 and the number of women aged 15 to 19, calculated for each municipality. We first applied multilevel models separately for 2000 and 2010 to test the cross-sectional association between income inequality and adolescent fertility. We then fitted longitudinal first-differences multilevel models to control for time-invariant effects. We also performed a sensitivity analysis to include only municipality with satisfactory birth record coverage. RESULTS: Our results indicate a consistent and positive association between income inequality and adolescent fertility. After controlling for per capita income, college access, youth homicide rate and adult fertility, higher income inequality was significantly associated with higher adolescent fertility for both 2000 and 2010. The longitudinal multilevel models found similar results. The sensitivity analysis indicated that the results for the association between income inequality and adolescent fertility were robust. Adult fertility was also significantly associated with adolescent fertility in the cross-sectional and longitudinal models. CONCLUSION: Income inequality is expected to be a leading concern for most countries in the near future. Our results suggest that changes in income inequality are positively and consistently associated with changes in adolescent fertility.
Assuntos
Coeficiente de Natalidade , Cidades , Renda/estatística & dados numéricos , Gravidez na Adolescência/estatística & dados numéricos , Adolescente , Brasil , Estudos Transversais , Feminino , Humanos , Análise Multinível , Gravidez , Fatores Socioeconômicos , Adulto JovemRESUMO
BACKGROUND: With the recent increase in the prevalence of mental disorders in developing countries, there is a growing interest in the study of its consequences. We examined the association of depression, anxiety and any mental disorders with incremental health expenditure, i.e. the linear increase in health expenditure associated with mental disorders, and lost days of normal activity. METHODS: We analyzed the results from a representative sample survey of residents of the Metropolitan Region of São Paulo (n = 2,920; São Paulo Megacity Mental Health Survey), part of the World Mental Health (WMH) Survey Initiative, coordinated by the World Health Organization and performed in 28 countries. The instrument used for obtaining the individual results, including the assessment of mental disorders, was the WMH version of the Composite International Diagnostic Interview 3.0 (WMH-CIDI 3.0) that generates psychiatric diagnoses according to the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV) criteria. Statistical analyses were performed by multilevel generalized least squares (GLS) regression models. Sociodemographic determinants such as income, age, education and marital status were included as controls. RESULTS: Depression, anxiety and any mental disorders were consistently associated with both incremental health expenditure and missing days of normal activity. Depression was associated with an incremental annual expenditure of R$308.28 (95% CI: R$194.05-R$422.50), or US$252.48 in terms of purchasing power parity (PPP). Anxiety and any mental disorders were associated with a lower, but also statistically significant, incremental annual expenditure (R$177.82, 95% CI: 79.68-275.97; and R$180.52, 95% CI: 91.13-269.92, or US$145.64 and US$147.85 in terms of PPP, respectively). Most of the incremental health costs associated with mental disorders came from medications. Depression was independently associated with higher incremental health expenditure than the two most prevalent chronic diseases found by the study (hypertension and diabetes). CONCLUSIONS: The fact that individuals with mental disorders had a consistent higher health expenditure is notable given the fact that Brazil has a universal free-of-charge healthcare and medication system. The results highlight the growing importance of mental disorders as a public health issue for developing countries.
Assuntos
Gastos em Saúde/estatística & dados numéricos , Transtornos Mentais/economia , Transtornos Mentais/epidemiologia , Saúde Mental/economia , Saúde Mental/estatística & dados numéricos , Absenteísmo , Atividades Cotidianas , Adulto , Idoso , Ansiedade/economia , Ansiedade/epidemiologia , Brasil/epidemiologia , Depressão/economia , Depressão/epidemiologia , Feminino , Inquéritos Epidemiológicos/estatística & dados numéricos , Humanos , Renda/estatística & dados numéricos , Masculino , Pessoa de Meia-Idade , Prevalência , População Urbana/estatística & dados numéricos , Adulto JovemRESUMO
OBJECTIVES: We calculated life expectancy at birth for Whites, Blacks, and mixed races in Brazil, and decomposed the differences by causes of death. METHODS: We used Ministry of Health death records and 2010 Census population data (190,755,799 residents and 1,136,947 deaths). We applied the Arriaga methodology to calculate decomposition of life expectancy by cause of death. We performed sensitivity analyses for underreporting of deaths, missing data, and numerator-denominator bias. RESULTS: Using standard life table methods, female life expectancy was highest for mixed races (78.80 years), followed by Whites (77.54 years), then Blacks (76.32 years). Male life expectancy was highest for Whites (71.10 years) followed closely by mixed races (71.08 years), and lower for Blacks (70.11 years). Homicides contributed the most to the relative life expectancy increase for Whites, and cancer decreased the gap. After adjustment for underreporting, missing data, and numerator-denominator bias, life expectancy was higher for Whites than for Blacks and mixed races. CONCLUSIONS: Despite wide socioeconomic differences between Whites and mixed races, standard life table methods showed that mixed races had higher life expectancy than Whites for women, and similar for men. With the increase of multiracial populations, measuring racial disparities in life expectancy will be a fast-growing challenge.
Assuntos
Disparidades nos Níveis de Saúde , Expectativa de Vida , Grupos Raciais/estatística & dados numéricos , Adolescente , Adulto , Idoso , Idoso de 80 Anos ou mais , População Negra/estatística & dados numéricos , Brasil/epidemiologia , Criança , Pré-Escolar , Feminino , Humanos , Lactente , Tábuas de Vida , Masculino , Pessoa de Meia-Idade , População Branca/estatística & dados numéricos , Adulto JovemRESUMO
INTRODUCTION: Chagas disease is a severe parasitic illness that is prevalent in Latin America and often goes unaddressed. Early detection and treatment are critical in preventing the progression of the illness and its associated life-threatening complications. In recent years, machine learning algorithms have emerged as powerful tools for disease prediction and diagnosis. METHODS: In this study, we developed machine learning algorithms to predict the risk of Chagas disease based on five general factors: age, gender, history of living in a mud or wooden house, history of being bitten by a triatomine bug, and family history of Chagas disease. We analyzed data from the Retrovirus Epidemiology Donor Study (REDS) to train five popular machine learning algorithms. The sample comprised 2,006 patients, divided into 75% for training and 25% for testing algorithm performance. We evaluated the model performance using precision, recall, and AUC-ROC metrics. RESULTS: The Adaboost algorithm yielded an AUC-ROC of 0.772, a precision of 0.199, and a recall of 0.612. We simulated the decision boundary using various thresholds and observed that in this dataset a threshold of 0.45 resulted in a 100% recall. This finding suggests that employing such a threshold could potentially save 22.5% of the cost associated with mass testing of Chagas disease. CONCLUSION: Our findings highlight the potential of applying machine learning to improve the sensitivity and effectiveness of Chagas disease diagnosis and prevention. Furthermore, we emphasize the importance of integrating socio-demographic and environmental factors into neglected disease prediction models to enhance their performance.
Assuntos
Doença de Chagas , Aprendizado de Máquina , População Rural , Humanos , Doença de Chagas/epidemiologia , Doença de Chagas/diagnóstico , Brasil/epidemiologia , Masculino , Feminino , Adulto , Pessoa de Meia-Idade , Adulto Jovem , Adolescente , Algoritmos , Criança , Fatores de Risco , Idoso , Pré-EscolarRESUMO
OBJECTIVES: We determined whether community-level income inequality was associated with mortality among a cohort of older adults in São Paulo, Brazil. METHODS: We analyzed the Health, Well-Being, and Aging (SABE) survey, a sample of community-dwelling older adults in São Paulo (2000-2007). We used survival analysis to examine the relationship between income inequality and risk for mortality among older individuals living in 49 districts of São Paulo. RESULTS: Compared with individuals living in the most equal districts (lowest Gini quintile), rates of mortality were higher for those living in the second (adjusted hazard ratio [AHR] = 1.44, 95% confidence interval [CI] = 0.87, 2.41), third (AHR = 1.96, 95% CI = 1.20, 3.20), fourth (AHR = 1.34, 95% CI = 0.81, 2.20), and fifth quintile (AHR = 1.74, 95% CI = 1.10, 2.74). When we imputed missing data and used poststratification weights, the adjusted hazard ratios for quintiles 2 through 5 were 1.72 (95% CI = 1.13, 2.63), 1.41 (95% CI = 0.99, 2.05), 1.13 (95% = 0.75, 1.70) and 1.30 (95% CI = 0.90, 1.89), respectively. CONCLUSIONS: We did not find a dose-response relationship between area-level income inequality and mortality. Our findings could be consistent with either a threshold association of income inequality and mortality or little overall association.
Assuntos
Disparidades nos Níveis de Saúde , Renda/estatística & dados numéricos , Mortalidade , Idoso , Brasil/epidemiologia , Feminino , Nível de Saúde , Humanos , Estudos Longitudinais , Masculino , Modelos de Riscos Proporcionais , Características de Residência/estatística & dados numéricos , Análise de SobrevidaRESUMO
Machine learning algorithms are being increasingly used in healthcare settings but their generalizability between different regions is still unknown. This study aims to identify the strategy that maximizes the predictive performance of identifying the risk of death by COVID-19 in different regions of a large and unequal country. This is a multicenter cohort study with data collected from patients with a positive RT-PCR test for COVID-19 from March to August 2020 (n = 8477) in 18 hospitals, covering all five Brazilian regions. Of all patients with a positive RT-PCR test during the period, 2356 (28%) died. Eight different strategies were used for training and evaluating the performance of three popular machine learning algorithms (extreme gradient boosting, lightGBM, and catboost). The strategies ranged from only using training data from a single hospital, up to aggregating patients by their geographic regions. The predictive performance of the algorithms was evaluated by the area under the ROC curve (AUROC) on the test set of each hospital. We found that the best overall predictive performances were obtained when using training data from the same hospital, which was the winning strategy for 11 (61%) of the 18 participating hospitals. In this study, the use of more patient data from other regions slightly decreased predictive performance. However, models trained in other hospitals still had acceptable performances and could be a solution while data for a specific hospital is being collected.
Assuntos
COVID-19 , Humanos , COVID-19/diagnóstico , COVID-19/epidemiologia , Estudos de Coortes , Algoritmos , Aprendizado de Máquina , Avaliação de Resultados em Cuidados de Saúde , Estudos RetrospectivosRESUMO
Objectives: Our aim was to test if machine learning algorithms can predict cancer mortality (CM) at an ecological level and use these results to identify statistically significant spatial clusters of excess cancer mortality (eCM). Methods: Age-standardized CM was extracted from the official databases of Brazil. Predictive features included sociodemographic and health coverage variables. Machine learning algorithms were selected and trained with 70% of the data, and the performance was tested with the remaining 30%. Clusters of eCM were identified using SatScan. Additionally, separate analyses were performed for the 10 most frequent cancer types. Results: The gradient boosting trees algorithm presented the highest coefficient of determination (R 2 = 0.66). For total cancer, all algorithms overlapped in the region of Bagé (27% eCM). For esophageal cancer, all algorithms overlapped in west Rio Grande do Sul (48%-96% eCM). The most significant cluster for stomach cancer was in Macapá (82% eCM). The most important variables were the percentage of the white population and residents with computers. Conclusion: We found consistent and well-defined geographic regions in Brazil with significantly higher than expected cancer mortality.
Assuntos
Neoplasias , Humanos , Brasil/epidemiologia , Aprendizado de Máquina , AlgoritmosRESUMO
BACKGROUND: After kidney transplantation (KTx), the graft can evolve from excellent immediate graft function (IGF) to total absence of function requiring dialysis. Recipients with IGF do not seem to benefit from using machine perfusion, an expensive procedure, in the long term when compared with cold storage. This study proposes to develop a prediction model for IGF in KTx deceased donor patients using machine learning algorithms. METHODS: Unsensitized recipients who received their first KTx deceased donor between January 1, 2010, and December 31, 2019, were classified according to the conduct of renal function after transplantation. Variables related to the donor, recipient, kidney preservation, and immunology were used. The patients were randomly divided into 2 groups: 70% were assigned to the training and 30% to the test group. Popular machine learning algorithms were used: eXtreme Gradient Boosting (XGBoost), Light Gradient Boosting Machine, Gradient Boosting classifier, Logistic Regression, CatBoost classifier, AdaBoost classifier, and Random Forest classifier. Comparative performance analysis on the test dataset was performed using the results of the AUC values, sensitivity, specificity, positive predictive value, negative predictive value, and F1 score. RESULTS: Of the 859 patients, 21.7% (n = 186) had IGF. The best predictive performance resulted from the eXtreme Gradient Boosting model (AUC, 0.78; 95% CI, 0.71-0.84; sensitivity, 0.64; specificity, 0.78). Five variables with the highest predictive value were identified. CONCLUSIONS: Our results indicated the possibility of creating a model for the prediction of IGF, enhancing the selection of patients who would benefit from an expensive treatment, as in the case of machine perfusion preservation.
Assuntos
Transplante de Rim , Humanos , Rim/fisiologia , Doadores de Tecidos , Valor Preditivo dos Testes , Aprendizado de MáquinaRESUMO
Importance: Access to routine dental care prevents advanced dental disease and improves oral and overall health. Identifying individuals at risk of foregoing preventive dental care can direct prevention efforts toward high-risk populations. Objective: To predict foregone preventive dental care among adults overall and in sociodemographic subgroups and to assess the algorithmic fairness. Design, Setting, and Participants: This prognostic study was a secondary analyses of longitudinal data from the US Medical Expenditure Panel Survey (MEPS) from 2016 to 2019, each with 2 years of follow-up. Participants included adults aged 18 years and older. Data analysis was performed from December 2022 to June 2023. Exposure: A total of 50 predictors, including demographic and socioeconomic characteristics, health conditions, behaviors, and health services use, were assessed. Main Outcomes and Measures: The outcome of interest was foregoing preventive dental care, defined as either cleaning, general examination, or an appointment with the dental hygienist, in the past year. Results: Among 32â¯234 participants, the mean (SD) age was 48.5 (18.2) years and 17â¯386 participants (53.9%) were female; 1935 participants (6.0%) were Asian, 5138 participants (15.9%) were Black, 7681 participants (23.8%) were Hispanic, 16â¯503 participants (51.2%) were White, and 977 participants (3.0%) identified as other (eg, American Indian and Alaska Native) or multiple racial or ethnic groups. There were 21â¯083 (65.4%) individuals who missed preventive dental care in the past year. The algorithms demonstrated high performance, achieving an area under the receiver operating characteristic curve (AUC) of 0.84 (95% CI, 0.84-0.85) in the overall population. While the full sample model performed similarly when applied to White individuals and older adults (AUC, 0.88; 95% CI, 0.87-0.90), there was a loss of performance for other subgroups. Removing the subgroup-sensitive predictors (ie, race and ethnicity, age, and income) did not impact model performance. Models stratified by race and ethnicity performed similarly or worse than the full model for all groups, with the lowest performance for individuals who identified as other or multiple racial groups (AUC, 0.76; 95% CI, 0.70-0.81). Previous pattern of dental visits, health care utilization, dental benefits, and sociodemographic characteristics were the highest contributing predictors to the models' performance. Conclusions and Relevance: Findings of this prognostic study using cohort data suggest that tree-based ensemble machine learning models could accurately predict adults at risk of foregoing preventive dental care and demonstrated bias against underrepresented sociodemographic groups. These results highlight the importance of evaluating model fairness during development and testing to avoid exacerbating existing biases.
Assuntos
Etnicidade , Grupos Raciais , Humanos , Idoso , Algoritmos , Aprendizado de Máquina , Assistência OdontológicaRESUMO
OBJETIVO: To describe the initial baseline results of a population-based study, as well as a protocol in order to evaluate the performance of different machine learning algorithms with the objective of predicting the demand for urgent and emergency services in a representative sample of adults from the urban area of Pelotas, Southern Brazil. METHODS: The study is entitled "Emergency department use and Artificial Intelligence in PELOTAS (RS) (EAI PELOTAS)" (https://wp.ufpel.edu.br/eaipelotas/). Between September and December 2021, a baseline was carried out with participants. A follow-up was planned to be conducted after 12 months in order to assess the use of urgent and emergency services in the last year. Afterwards, machine learning algorithms will be tested to predict the use of urgent and emergency services over one year. RESULTS: In total, 5,722 participants answered the survey, mostly females (66.8%), with an average age of 50.3 years. The mean number of household people was 2.6. Most of the sample has white skin color and incomplete elementary school or less. Around 30% of the sample has obesity, 14% diabetes, and 39% hypertension. CONCLUSION: The present paper presented a protocol describing the steps that were and will be taken to produce a model capable of predicting the demand for urgent and emergency services in one year among residents of Pelotas, in Rio Grande do Sul state.