A random forest model to predict heatstroke occurrence for heatwave in China

Chasing the heat: Unraveling urban hyperlocal air temperature mapping with mobile sensing and machine learning

Many cities face unprecedented high temperatures with increasing extreme events. Heatwaves pose significant health risks, including cardiovascular diseases, heatstroke, and dehydration. Mapping urban near-surface air temperature (Tair) is crucial for understanding thermal exposure and addressing climate change. Previous studies relied on satellite-derived land surface temperature (LST) and stationary monitoring, but high spaio-temporal Tair mapping is still a challenge. This study optimized a mobile sensing scheme using an electric bicycle platform with environmental and image sensors, and deep learning captured local-scale urban factors. A spatio-temporal data fusion model that consisted of three parts, temporal trend extraction, locality analysis, and neighborhood effect analysis, generated hyperlocal Tair maps. The Results from Beijing demonstrated the effectiveness of the framework, achieving the lowest MAE of 0.02 °C. Optimized data collection and the new model achieved accurate temperature predictions and thermal exposure assessment. Efficiency enhanced sensing strategy was also proposed. The study highlights local-scale factors and spatio-temporal dependencies in addressing heatwaves and climate change impacts in urban areas.

A new circa 2007 biomass map for China differs significantly from existing maps

The forest area of China is the fifth largest of any country, and unlike in many other countries, in recent decades its area has been increasing. However, there are substantial differences in estimates of the amount of carbon this forest contains, ranging from 3.92 to 17.02 Pg C for circa 2007. This makes it unclear how the changes in China's forest area contribute to the global carbon cycle. We generate a circa 2007 aboveground biomass (AGB) map at a resolution of 50 m using optical, radar and LiDAR satellite data. Our estimates of total carbon stored in the forest in China was 9.52 Pg C, with an average forest AGB of 104 Mg ha-1. Compared with three existing AGB maps, our AGB map showed better correlation with a distributed set of forest inventory plots. In addition, our high resolution AGB map provided more details on spatial distribution of forest AGB, and is likely to help understand the carbon storage changes in China's forest.

Revisiting the importance of temperature, weather and air pollution variables in heat-mortality relationships with machine learning

Extreme heat events have significant health impacts that need to be adequately quantified in the context of climate change. Traditionally, heat-health association methods have relied on statistical models using a single air temperature index, without considering other heat-related variables that may influence the relationship and their potentially complex interactions. This study aims to introduce and compare different machine learning (ML) models, which naturally consider interactions between predictors and non-linearities, to re-examine the importance of temperature, weather and air pollution predictors in modeling the heat-mortality relationship. ML approaches based on tree ensembles and neural networks, as well as non-linear statistical models, were used to model the heat-mortality relationship in the two most populated metropolitan areas of the province of Quebec, Canada. The models were calibrated using a comprehensive database of heat-related predictors including various lagged temperature indices, temperature variations, meteorological and air pollution variables. Performance was evaluated based on out-of-sample summer mortality predictions. For the two studied regions, models relying only on lagged temperature indices performed better, or equally well, than models considering more heat-related predictors such as temperature variations, weather and air pollution variables. The temperature index with the best performance differed by region, but both mean temperature and humidex were among the best indices. In terms of modeling approaches, non-linear statistical models were as competent as more advanced ML models for predicting out-of-sample summer mortality. This research validated the current use of non-linear statistical models with the appropriate lagged temperature index to model the heat-mortality relationship. Although ML models have not improved the performance of all-cause mortality modeling, these approaches should continue to be explored, particularly for other health effects that may be more directly linked to heat exposure and, in the future, when more data become available.

Socio-Environmental Determinants of Mental and Behavioral Disorders in Youth: A Machine Learning Approach

Growing evidence indicates that extreme environmental conditions in summer months have an adverse impact on mental and behavioral disorders (MBD), but there is limited research looking at youth populations. The objective of this study was to apply machine learning approaches to identify key variables that predict MBD-related emergency room (ER) visits in youths in select North Carolina cities among adolescent populations. Daily MBD-related ER visits, which totaled over 42,000 records, were paired with daily environmental conditions, as well as sociodemographic variables to determine if certain conditions lead to higher vulnerability to exacerbated mental health disorders. Four machine learning models (i.e., generalized linear model, generalized additive model, extreme gradient boosting, random forest) were used to assess the predictive performance of multiple environmental and sociodemographic variables on MBD-related ER visits for all cities. The best-performing machine learning model was then applied to each of the six individual cities. As a subanalysis, a distributed lag nonlinear model was used to confirm results. In the all cities scenario, sociodemographic variables contributed the greatest to the overall MBD prediction. In the individual cities scenario, four cities had a 24-hr difference in the maximum temperature, and two of the cities had a 24-hr difference in the minimum temperature, maximum temperature, or Normalized Difference Vegetation Index as a leading predictor of MBD ER visits. Results can inform the use of machine learning models for predicting MBD during high-temperature events and identify variables that affect youth MBD responses during these events.

Exploring patient medication adherence and data mining methods in clinical big data: A contemporary review

BACKGROUND

Increasingly, patient medication adherence data are being consolidated from claims databases and electronic health records (EHRs). Such databases offer an indirect avenue to gauge medication adherence in our data-rich healthcare milieu. The surge in data accessibility, coupled with the pressing need for its conversion to actionable insights, has spotlighted data mining, with machine learning (ML) emerging as a pivotal technique. Nonadherence poses heightened health risks and escalates medical costs. This paper elucidates the synergistic interaction between medical database mining for medication adherence and the role of ML in fostering knowledge discovery.

METHODS

We conducted a comprehensive review of EHR applications in the realm of medication adherence, leveraging ML techniques. We expounded on the evolution and structure of medical databases pertinent to medication adherence and harnessed both supervised and unsupervised ML paradigms to delve into adherence and its ramifications.

RESULTS

Our study underscores the applications of medical databases and ML, encompassing both supervised and unsupervised learning, for medication adherence in clinical big data. Databases like SEER and NHANES, often underutilized due to their intricacies, have gained prominence. Employing ML to excavate patient medication logs from these databases facilitates adherence analysis. Such findings are pivotal for clinical decision-making, risk stratification, and scholarly pursuits, aiming to elevate healthcare quality.

CONCLUSION

Advanced data mining in the era of big data has revolutionized medication adherence research, thereby enhancing patient care. Emphasizing bespoke interventions and research could herald transformative shifts in therapeutic modalities.

Temporal and spatial characteristics of high temperatures, heat waves, and population distribution risk in China from 1951 to 2019

Understanding the relationships between high temperatures (HT) and heat waves (HW) is vital for enhancing human health, especially in areas with dense population. This paper analyzes the temporal and spatial characteristics of different HT and HW intensities, their spatial influence, and the population distribution risk at different HW intensities for 844 meteorological stations between 1951 and 2019. The results indicate that (1) HT and extreme temperature (ET) days are symmetrically distributed along the Huhuanyong Line, from southeast to northwest China. The times, days, and accumulated temperatures of HW, the times, days, and accumulated temperature of strong heat waves (SHW), and the times, days, and accumulated temperature of extreme heat waves (EHW) were distributed similarly; (2) with the increase in high temperatures or heat waves from HT to ET or from HW to SHW, the proportion of stations with an upward trend was always greater in China, while stations with a downward trend were mainly located in the North China Plain and Huai River Basin. For HW, SHW, and EHW, the increasing range of times and days were less than the accumulated temperatures; (3) between 1990 and 2019, there was an expansion of the HW and SHW distribution area with an annual average of more than 10 days, and the EHW distribution area with an annual average of more than 3 days. Moreover, the number of people exposed to HW, SHW, and EHW also increased during this period; and (4) considering the population distribution characteristics and the regional HT and HW characteristics, society needs to form regional adaptation actions for different HT and HW intensities.

Effects of heatwave features on machine-learning-based heat-related ambulance calls prediction models in Japan

Researchers agree that there is substantial evidence of an increasing trend in both the frequency and duration of extreme temperature events. Increasing extreme temperature events will place more pressure on public health and emergency medical resources, and societies will need to find effective and reliable solutions to adapt to hotter summers. This study developed an effective method to predict the number of daily heat-related ambulance calls. Both national- and regional-level models were developed to evaluate the performance of machine-learning-based methods on heat-related ambulance call prediction. The national model showed a high prediction accuracy and can be applied over most regions, while the regional model showed extremely high prediction accuracy in each corresponding region and reliable accuracy in special cases. We found that the introduction of heatwave features, including accumulated heat stress, heat acclimatization, and optimal temperature, significantly improved prediction accuracy. The adjusted coefficient of determination (adjusted R²) of the national model improved from 0.9061 to 0.9659 by including these features, and the adjusted R² of the regional model also improved from 0.9102 to 0.9860. Furthermore, we used five bias-corrected global climate models (GCMs) to forecast the total number of summer heat-related ambulance calls under three different future climate scenarios nationally and regionally. Our analysis demonstrated that, at the end of the 21st century, the total number of heat-related ambulance calls in Japan will reach approximately 250,000 per year (nearly four times the current amount) under SSP-5.85. Our results suggest that disaster management agencies can use this highly accurate model to forecast potential high emergency medical resource burden caused by extreme heat events, allowing them to raise and improve public awareness and prepare countermeasures in advance. The method proposed in Japan in this paper can be applied to other countries that have relevant data and weather information systems.

Estimation of the number of heat illness patients in eight metropolitan prefectures of Japan: Correlation with ambient temperature and computed thermophysiological responses

The number of patients with heat illness transported by ambulance has been gradually increasing due to global warming. In intense heat waves, it is crucial to accurately estimate the number of cases with heat illness for management of medical resources. Ambient temperature is an essential factor with respect to the number of patients with heat illness, although thermophysiological response is a more relevant factor with respect to causing symptoms. In this study, we computed daily maximum core temperature increase and daily total amount of sweating in a test subject using a large-scale, integrated computational method considering the time course of actual ambient conditions as input. The correlation between the number of transported people and their thermophysiological temperature is evaluated in addition to conventional ambient temperature. With the exception of one prefecture, which features a different Köppen climate classification, the number of transported people in the remaining prefectures, with a Köppen climate classification of Cfa, are well estimated using either ambient temperature or computed core temperature increase and daily amount of sweating. For estimation using ambient temperature, an additional two parameters were needed to obtain comparable accuracy. Even using ambient temperature, the number of transported people can be estimated if the parameters are carefully chosen. This finding is practically useful for the management of ambulance allocation on hot days as well as public enlightenment.

Nexus of heat-vulnerable chronic diseases and heatwave mediated through tri-environmental interactions: A nationwide fine-grained study in Australia

The warming trend over recent decades has already contributed to the increased prevalence of heat-vulnerable chronic diseases in many regions of the world. However, understanding the relationship between heat-vulnerable chronic diseases and heatwaves remains incomplete due to the complexity of such a relationship mingling with human society, urban and natural environments. Our study extends the Social Ecological Theory by constructing a tri-environmental conceptual framework (i.e., across social, built, and natural environments) and contributes to the first nationwide study of the relationship between heat-vulnerable chronic diseases and heatwaves in Australia. We utilize the random forest regression model to explore the importance of heatwaves and 48 tri-environmental variables that contribute to the prevalence of six types of heat-vulnerable diseases. We further apply the local interpretable model-agnostic explanations and the accumulated local effects analysis to interpret how the heat-disease nexus is mediated through tri-environments and varied across urban and rural space. The overall effect of heatwaves on diseases varies across disease types and geographical contexts (latitudes; inland versus coast). The local heat-disease nexus follows a J-shape function-becoming sharply positive after a certain threshold of heatwaves-reflecting that people with the onset of different diseases have various sensitivity and tolerance to heatwaves. However, such effects are relatively marginal compared to tri-environmental variables. We propose a number of policy implications on reducing urban-rural disparity in healthcare access and service distribution, delineating areas, and identifying the variations of sensitivity to heatwaves across urban/rural space and disease types. Our conceptual framework can be further applied to examine the relationship between other environmental problems and health outcomes.

Thermal comfort differences between urban villages and formal settlements in Chinese developing cities: A case study in Shenzhen

Rapid urbanization has changed the urban spatial form, which directly leads to the emergence of urban informal settlements, and the impact on ecological environment is manifested as the obvious deterioration of urban thermal environment. The thermal environment of informal settlements, which are called urban villages in China, is seriously deteriorated. In the process of urban renewal, we should pay attention to the thermal environment effect of urban villages and promote the sustainable development of cities. However, at present there are few studies on the differences of thermal comfort among urban settlements. Taking Shenzhen as an example city, this paper distinguished several scopes such as urban villages, formal settlements and non-urban areas, then analyzed the pattern characteristics of urban thermal comfort by using the Modified Temperature and Humidity Index (MTHI), and finally explored the spatial relationship between thermal comfort and various environmental factors through spatial regression models. The results show that (1) thermal comfort has significant spatial autocorrelation and the thermal environment centers are clustered in built-up areas of Shenzhen city. The overall MTHI of urban villages is relatively the highest, and the dominant thermal comfort level in summer is sultry and hot. (2) According to the Spatial Error Model, the green space coverage represented by NDVI has the strongest mitigation effect on urban thermal comfort, and the building density has an obvious aggravating effect on the muggy environment. Both of them have more obvious effects on the thermal comfort of informal settlements. (3) The current thermal environment of urban settlements cannot be ignored, especially in urban villages. In the process of urban renewal, heat dissipation should be considered emphatically. Increasing urban green area and decreasing building density will help to improve urban thermal environment. The study can provide suggestions for urban renewal from the perspective of improving thermal environment.

Environmental variable importance for under-five mortality in Malaysia: A random forest approach

BACKGROUND

Environmental factors have been associated with adverse health effects in epidemiological studies. The main exposure variable is usually determined via prior knowledge or statistical methods. It may be challenging when evidence is scarce to support prior knowledge, or to address collinearity issues using statistical methods. This study aimed to investigate the importance level of environmental variables for the under-five mortality in Malaysia via random forest approach.

METHOD

We applied a conditional permutation importance via a random forest (CPI-RF) approach to evaluate the relative importance of the weather- and air pollution-related environmental factors on daily under-five mortality in Malaysia. This study spanned from January 1, 2014 to December 31, 2016. In data preparation, deviation mortality counts were derived through a generalized additive model, adjusting for long-term trend and seasonality. Analyses were conducted considering mortality causes (all-cause, natural-cause, or external-cause) and data structures (continuous, categorical, or all types [i.e., include all variables of continuous type and all variables of categorical type]). The main analysis comprised of two stages. In Stage 1, Boruta selection was applied for preliminary screening to remove highly unimportant variables. In Stage 2, the retained variables from Boruta were used in the CPI-RF analysis. The final importance value was obtained as an average value from a 10-fold cross-validation.

RESULT

Some heat-related variables (maximum temperature, heat wave), temperature variability, and haze-related variables (PM10, PM10-derived haze index, PM10- and fire-derived haze index, fire hotspot) were among the prominent variables associated with under-five mortality in Malaysia. The important variables were consistent for all- and natural-cause mortality and sensitivity analyses. However, different most important variables were observed between natural- and external-cause under-five mortality.

CONCLUSION

Heat-related variables, temperature variability, and haze-related variables were consistently prominent for all- and natural-cause under-five mortalities, but not for external-cause.

Web-Based Data to Quantify Meteorological and Geographical Effects on Heat Stroke: Case Study in China

Heat stroke is a serious heat-related health outcome that can eventually lead to death. Due to the poor accessibility of heat stroke data, the large-scale relationship between heat stroke and meteorological factors is still unclear. This work aims to clarify the potential relationship between meteorological variables and heat stroke, and quantify the meteorological threshold that affected the severity of heat stroke. We collected daily heat stroke search index (HSSI) and meteorological data for the period 2013-2020 in 333 Chinese cities to analyze the relationship between meteorological variables and HSSI using correlation analysis and Random forest (RF) model. Temperature and relative humidity (RH) accounted for 62% and 9% of the changes of HSSI, respectively. In China, cases of heat stroke may start to occur when temperature exceeds 36°C and RH exceeds 58%. This threshold was 34.5°C and 79% in the north of China, and 36°C and 48% in the south of China. Compared to RH, the threshold of temperature showed a more evident difference affected by altitude and distance from the ocean, which was 35.5°C in inland cities and 36.5°C in coastal cities; 35.5°C in high-altitude cities and 36°C in low-altitude cities. Our findings provide a possible way to analyze the interaction effect of meteorological variables on heat-related illnesses, and emphasizes the effects of geographical environment. The meteorological threshold quantified in this research can also support policymaker to establish a better meteorological warning system for public health.

Machine Learning Analyzed Weather Conditions as an Effective Means in the Predicting of Acute Coronary Syndrome Prevalence

Background

The prediction of the number of acute coronary syndromes (ACSs) based on the weather conditions in the individual climate zones is not effective. We sought to investigate whether an artificial intelligence system might be useful in this prediction.

Methods

Between 2008 and 2018, a total of 105,934 patients with ACS were hospitalized in Lesser Poland Province, one covered by two meteorological stations. The predicted daily number of ACS has been estimated with the Random Forest machine learning system based on air temperature (°C), air pressure (hPa), dew point temperature (Td) (°C), relative humidity (RH) (%), wind speed (m/s), and precipitation (mm) and their daily extremes and ranges derived from the day of ACS and from 6 days before ACS.

Results

Of 840 pairwise comparisons between individual weather parameters and the number of ACS, 128 (15.2%) were significant but weak with the correlation coefficients ranged from −0.16 to 0.16. None of weather parameters correlated with the number of ACS in all the seasons and stations. The number of ACS was higher in warm front days vs. days without any front [40 (29–50) vs. 38 (27–48), respectively, P < 0.05]. The correlation between the predicted and observed daily number of ACS derived from machine learning was 0.82 with 95% CI of 0.80–0.84 (P < 0.001). The greatest importance for machine learning (range 0–1.0) among the parameters reached Td daily range with 1.00, pressure daily range with 0.875, pressure maximum daily range with 0.864, and RH maximum daily range with 0.853, whereas among the clinical parameters reached hypertension daily range with 1.00 and diabetes mellitus daily range with 0.28. For individual seasons and meteorological stations, the correlations between the predicted and observed number of ACS have ranged for spring from 0.73 to 0.77 (95% CI 0.68–0.82), for summer from 0.72 to 0.76 (95% CI 0.66–0.81), for autumn from 0.72 to 0.83 (95% CI 0.67–0.87), and for winter from 0.76 to 0.79 (95% CI 0.71–0.83) (P < 0.001 for each).

Conclusion

The weather parameters have proven useful in predicting the prevalence of ACS in a temperate climate zone for all the seasons, if analyzed with an artificial intelligence system. Simultaneously, the analysis of individual weather parameters or frontal scenarios has provided only weak univariate relationships. These findings will require validation in other climatic zones.

Effective Macrosomia Prediction Using Random Forest Algorithm

(1) Background: Macrosomia is prevalent in China and worldwide. The current method of predicting macrosomia is ultrasonography. We aimed to develop new predictive models for recognizing macrosomia using a random forest model to improve the sensitivity and specificity of macrosomia prediction; (2) Methods: Based on the Shandong Multi-Center Healthcare Big Data Platform, we collected the prenatal examination and delivery data from June 2017 to May 2018 in Jinan, including the macrosomia and normal-weight newborns. We constructed a random forest model and a logistic regression model for predicting macrosomia. We compared the validity and predictive value of these two methods and the traditional method; (3) Results: 405 macrosomia cases and 3855 normal-weight newborns fit the selection criteria and 405 pairs of macrosomia and control cases were brought into the random forest model and logistic regression model. On the basis of the average decrease of the Gini coefficient, the order of influencing factors was: interspinal diameter, transverse outlet, intercristal diameter, sacral external diameter, pre-pregnancy body mass index, age, the number of pregnancies, and the parity. The sensitivity, specificity, and area under curve were 91.7%, 91.7%, and 95.3% for the random forest model, and 56.2%, 82.6%, and 72.0% for logistic regression model, respectively; the sensitivity and specificity were 29.6% and 97.5% for the ultrasound; (4) Conclusions: A random forest model based on the maternal information can be used to predict macrosomia accurately during pregnancy, which provides a scientific basis for developing rapid screening and diagnosis tools for macrosomia.

Soil Erosion Assessment of Alpine Grassland in the Source Park of the Yellow River on the Qinghai-Tibetan Plateau, China

The source park of the Yellow River (SPYR), as a vital ecological shelter on the Qinghai-Tibetan Plateau, is suffering different degrees of degradation and desertification, resulting in soil erosion in recent decades. Therefore, studying the mechanism, influencing factors and current situation of soil erosion in the alpine grassland ecosystems of the SPYR are significant for protecting the ecological and productive functions. Based on the 137Cs element tracing technique and machine learning algorithms, five strategic variable selection algorithms based on machine learning algorithms are used to identify the minimal optimal set and analyze the main factors that influence soil erosion in the SPYR. The optimal model for estimating soil erosion in the SPYR is obtained by comparisons model outputs between the RUSLE and machine learning algorithms combined with variable selection models. We identify the spatial distribution pattern of soil erosion in the study area by the optimal model. The results indicated that: (1) A comprehensive set of variables is more objective than the RUSLE model. In terms of verification accuracy, the simulated annealing -Cubist model (R = 0.67, RMSD = 1,368 t km–2⋅a–1) simulation results represents the best while the RUSLE model (R = 0.49, RMSD = 1,769 t⋅km–2⋅a–1) goes on the worst. (2) The soil erosion is more severe in the north than the southeast of the SPYR. The average erosion modulus is 6,460.95 t⋅km–2⋅a–1 and roughly 99% of the survey region has an intensive erosion modulus (5,000–8,000 t⋅km–2⋅a–1). (3) Total erosion loss is relatively 8.45⋅108 t⋅a–1 in the SPYR, which is commonly 12.64 times greater than the allowable soil erosion loss. The economic monetization of SOC loss caused by soil erosion in the entire research area was almost $47.90 billion in 2014. These results will help provide scientific evidences not only for farmers and herdsmen but also for environmental science managers and administrators. In addition, a new ecological policy recommendation was proposed to balance grassland protection and animal husbandry economic production based on the value of soil erosion reclassification.

Remote Sensing-Based Quantification of the Summer Maize Yield Gap Induced by Suboptimum Sowing Dates over North China Plain

Estimating yield potential (Yp) and quantifying the contribution of suboptimum field managements to the yield gap (Yg) of crops are important for improving crop yield effectively. However, achieving this goal on a regional scale remains difficult because of challenges in collecting field management information. In this study, we retrieved crop management information (i.e., emerging stage information and a surrogate of sowing date (SDT)) from a remote sensing (RS) vegetation index time series. Then, we developed a new approach to quantify maize Yp, total Yg, and the suboptimum SDT-induced Yg (Yg0) using a process-based RS-driven crop yield model for maize (PRYM–Maize), which was developed in our previous study. PRYM–Maize and the newly developed method were used over the North China Plain (NCP) to estimate Ya, Yp, Yg, and Yg0 of summer maize. Results showed that PRYM–Maize outputs reasonable estimates for maize yield over the NCP, with correlations and root mean standard deviation of 0.49 ± 0.24 and 0.88 ± 0.14 t hm−2, respectively, for modeled annual maize yields versus the reference value for each year over the period 2010 to 2015 on a city level. Yp estimated using our new method can reasonably capture the spatial variations in site-level estimates from crop growth models in previous literature. The mean annual regional Yp of 2010–2015 was estimated to be 11.99 t hm−2, and a Yg value of 5.4 t hm−2 was found between Yp and Ya on a regional scale. An estimated 29–42% of regional Yg in each year (2010–2015) was induced by suboptimum SDT. Results also show that not all Yg0 was persistent over time. Future studies using high spatial-resolution RS images to disaggregate Yg0 into persistent and non-persistent components on a small scale are required to increase maize yield over the NCP.

Heatstroke predictions by machine learning, weather information, and an all-population registry for 12-hour heatstroke alerts

This study aims to develop and validate prediction models for the number of all heatstroke cases, and heatstrokes of hospital admission and death cases per city per 12 h, using multiple weather information and a population-based database for heatstroke patients in 16 Japanese cities (corresponding to around a 10,000,000 population size). In the testing dataset, mean absolute percentage error of generalized linear models with wet bulb globe temperature as the only predictor and the optimal models, respectively, are 43.0% and 14.8% for spikes in the number of all heatstroke cases, and 37.7% and 10.6% for spikes in the number of heatstrokes of hospital admission and death cases. The optimal models predict the spikes in the number of heatstrokes well by machine learning methods including non-linear multivariable predictors and/or under-sampling and bagging. Here, we develop prediction models whose predictive performances are high enough to be implemented in public health settings.

In the context of climate change, heatstroke is expected to become an increasingly relevant public health concern. Here, the authors develop and validate prediction models for the number of all heatstroke cases in different cities in Japan.

Machine learning-based mortality prediction model for heat-related illness

In this study, we aimed to develop and validate a machine learning-based mortality prediction model for hospitalized heat-related illness patients. After 2393 hospitalized patients were extracted from a multicentered heat-related illness registry in Japan, subjects were divided into the training set for development (n = 1516, data from 2014, 2017–2019) and the test set (n = 877, data from 2020) for validation. Twenty-four variables including characteristics of patients, vital signs, and laboratory test data at hospital arrival were trained as predictor features for machine learning. The outcome was death during hospital stay. In validation, the developed machine learning models (logistic regression, support vector machine, random forest, XGBoost) demonstrated favorable performance for outcome prediction with significantly increased values of the area under the precision-recall curve (AUPR) of 0.415 [95% confidence interval (CI) 0.336–0.494], 0.395 [CI 0.318–0.472], 0.426 [CI 0.346–0.506], and 0.528 [CI 0.442–0.614], respectively, compared to that of the conventional acute physiology and chronic health evaluation (APACHE)-II score of 0.287 [CI 0.222–0.351] as a reference standard. The area under the receiver operating characteristic curve (AUROC) values were also high over 0.92 in all models, although there were no statistical differences compared to APACHE-II. This is the first demonstration of the potential of machine learning-based mortality prediction models for heat-related illnesses.

Forecast and early warning of hand, foot, and mouth disease based on meteorological factors: Evidence from a multicity study of 11 meteorological geographical divisions in mainland China

BACKGROUND

Hand, foot, and mouth disease (HFMD) is a significant public health issue in China. Early warning and forecasting are one of the most cost-effective ways for HFMD control and prevention. However, relevant research is limited, especially in China with a large population and diverse climatic characteristics. This study aims to identify local specific HFMD epidemic thresholds and construct a weather-based early warning model for HFMD control and prevention across China.

METHODS

Monthly notified HFMD cases and meteorological data for 22 cities selected from different climate zones from 2014 to 2018 were extracted from the National Notifiable Disease Surveillance System and the Meteorological Data Sharing Service System, respectively. A generalized additive model (GAM) based on meteorological factors was conducted to forecast HFMD epidemics. The receiver operator characteristic curve (ROC) was generated to determine the value of optimal warning threshold.

RESULTS

The developed model was solid in forecasting the epidemic of HFMD with all R square (R²) in the 22 cities above 85%, and mean absolute percentage error (MAPE) less than 1%. The warning thresholds varied by cities with the highest threshold observed in Shenzhen (n = 7195) and the lowest threshold in Liaoyang (n = 12). The areas under the curve (AUC) was greater than 0.9 for all regions, indicating a satisfied discriminating ability in epidemics detection.

CONCLUSIONS

The weather-based HFMD forecasting and early warning model we developed for different climate zones provides needed information on occurrence time and size of HFMD epidemics. An effective early warning system for HFMD could provide sufficient time for local authorities to implement timely interventions to minimize the HFMD morbidity and mortality.

Heatwave Damage Prediction Using Random Forest Model in Korea

Climate change increases the frequency and intensity of heatwaves, causing significant human and material losses every year. Big data, whose volumes are rapidly increasing, are expected to be used for preemptive responses. However, human cognitive abilities are limited, which can lead to ineffective decision making during disaster responses when artificial intelligence-based analysis models are not employed. Existing prediction models have limitations with regard to their validation, and most models focus only on heat-associated deaths. In this study, a random forest model was developed for the weekly prediction of heat-related damages on the basis of four years (2015–2018) of statistical, meteorological, and floating population data from South Korea. The model was evaluated through comparisons with other traditional regression models in terms of mean absolute error, root mean squared error, root mean squared logarithmic error, and coefficient of determination (R2). In a comparative analysis with observed values, the proposed model showed an R2 value of 0.804. The results show that the proposed model outperforms existing models. They also show that the floating population variable collected from mobile global positioning systems contributes more to predictions than the aggregate population variable.