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

Ogata, Soshiro; Takegami, Misa; Ozaki, Taira; Nakashima, Takahiro; Onozuka, Daisuke; Murata, Shunsuke; Nakaoku, Yuriko; Suzuki, Koyu; Hagihara, Akihito; Noguchi, Teruo; Iihara, Koji; Kitazume, Keiichi; Morioka, Tohru; Yamazaki, Shin; Yoshida, Takahiro; Yamagata, Yoshiki; Nishimura, Kunihiro

doi:10.1038/s41467-021-24823-0

Cited by 30 publications

(37 citation statements)

References 18 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…As mentioned in the introduction, published studies using machine learning methods for heatstroke prediction modelling have already been widely recognized ( Ogata et al, 2021 ; Ikeda & Kusaka, 2021 ). The aim of the present study was to generate a longer-term baseline scenario for heatstroke emergency transportation in Tokyo, using a dataset of climate change scenarios and possibly employing a more simplistic and yet tractable approach.…”

Section: Discussionmentioning

confidence: 99%

“…In Japan, an epidemiological study was conducted to predict the prevalence of heatstroke among elderly patients in an indoor environment ( Kodera et al, 2019 ). Moreover, machine learning has been employed to attempt to better predict such events over the course of time ( Ogata et al, 2021 ; Ikeda & Kusaka, 2021 ). Nevertheless, predictive scenarios over the long-term future have yet to be formulated.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Baseline scenarios of heat-related ambulance transportations under climate change in Tokyo, Japan

Fujimoto

Nishiura

2022

PeerJ

View full text Add to dashboard Cite

Background Predictive scenarios of heatstroke over the long-term future have yet to be formulated. The purpose of the present study was to generate baseline scenarios of heat-related ambulance transportations using climate change scenario datasets in Tokyo, Japan. Methods Data on the number of heat-related ambulance transportations in Tokyo from 2015 to 2019 were examined, and the relationship between the risk of heat-related ambulance transportations and the daily maximum wet-bulb globe temperature (WBGT) was modeled using three simple dose–response models. To quantify the risk of heatstroke, future climatological variables were then retrieved to compute the WBGT up to the year 2100 from climate change scenarios (i.e., RCP2.6, RCP4.5, and RCP8.5) using two scenario models. The predicted risk of heat-related ambulance transportations was embedded onto the future age-specific projected population. Results The proportion of the number of days with a WBGT above 28°C is predicted to increase every five years by 0.16% for RCP2.6, 0.31% for RCP4.5, and 0.68% for RCP8.5. In 2100, compared with 2000, the number of heat-related ambulance transportations is predicted to be more than three times greater among people aged 0–64 years and six times greater among people aged 65 years or older. The variance of the heatstroke risk becomes greater as the WBGT increases. Conclusions The increased risk of heatstroke for the long-term future was demonstrated using a simple statistical approach. Even with the RCP2.6 scenario, with the mildest impact of global warming, the risk of heatstroke is expected to increase. The future course of heatstroke predicted by our approach acts as a baseline for future studies.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Baseline scenarios of heat-related ambulance transportations under climate change in Tokyo, Japan

Fujimoto

Nishiura

2022

PeerJ

View full text Add to dashboard Cite

show abstract

“…Although heatstroke patient prediction models have been proposed in previous studies, they are not robust because of the restricted training data. For example, Ogata et al [20]. created a model with excellent accuracy, but it was not robust to training data because it used only one test dataset, 2018, and a xed three-year period from 2015.…”

Section: Discussionmentioning

confidence: 99%

“…By comparing the errors of each method, we identi ed factors that in uence the errors in the training data. As shown in the previous studies, the end of the rainy season, midsummer days (daily maximum temperature of 30°C or higher), extremely hot days (35°C or higher), and tropical nights (daily minimum temperature of 25°C or higher) were considered in uential factors for heatstroke patients [20,40]. We attempted a linear regression analysis using these factors as explanatory variables and RMSE as the objective variable to see if these factors are also determinants of model error.…”

Section: -7 Analyses Of In Uential Factorsmentioning

confidence: 99%

Comparative evaluation of accuracy of the multiple methods of heatstroke risk model including time-stratified case-crossover analysis

Nemoto

Hirabayashi²,

Sato

et al. 2022

Preprint

View full text Add to dashboard Cite

Background High temperatures in urban areas owing to climate change and urban heat islands have led to an increase in the number of heatstroke patients. To prevent heatstroke, accurate heatstroke patient prediction model should be used to predict and alert people to their risk. However, most previous models have not tested sufficient training data, although susceptibility to heatstroke is likely to be dependent on year-wise trends and is susceptible to training data. We investigated an accurate heatstroke risk model that is robust to the training data. By examining the factors affecting the accuracy and trade-off between the quantity and quality of adding old data to training data, the method of selecting training data for constructing an accurate model was also investigated. Method We compared the accuracies of three methods: multiple regression analysis (MR), generalized additive model (GAM), and time-stratified case-crossover analysis (TC). The training data for all combinations from 2012 were tested. By comparing the errors of each method, we identified the error influencing factors in the training data. Results The TC errors were the smallest (p < 0.005) and much less sensitive to the training data than others. Best accuracy odds ratios were 1.41–1.44. The error was significantly larger when the number of extremely hot days differed between the training and test data in MR and GAM (p < 0.01, p < 0.05). All three methods tended to increase accuracy up to a certain point and decrease from the middle of the year when adding past years retroactively from the most recent year in training data. Conclusions By using the odds ratios produced by TC with low sensitivity to training data, we can develop highly accurate heatstroke risk model that are robust to training data, which has not been possible with previous models. If data are to be included continuously, a model constructed from three or four years of data from the latest one is the most accurate.

show abstract

“…24−26 The under-sampling method was implemented to balance the numbers of positive and negative substances, which removed some samples in the label, with more substances to assure a balanced distribution. 27 Feature Extraction. The RDKit molecular descriptors representing chemical structures were employed as features for all substances.…”

Section: ■ Materials and Methodsmentioning

confidence: 99%

Machine Learning-Based Models with High Accuracy and Broad Applicability Domains for Screening PMT/vPvM Substances

Gao

Shen

et al. 2022

Environ. Sci. Technol.

View full text Add to dashboard Cite

Persistent, mobile, and toxic (PMT) substances and very persistent and very mobile (vPvM) substances can transport over long distances from various sources, increasing the public health risk. A rapid and high-throughput screening of PMT/vPvM substances is thus warranted to the risk prevention and mitigation measures. Herein, we construct a machine learning-based screening system integrated with five models for high-throughput classification of PMT/vPvM substances. The models are constructed with 44 971 substances by conventional learning, deep learning, and ensemble learning algorithms, among which, LightGBM and XGBoost outperform other algorithms with metrics exceeding 0.900. Good model interpretability is achieved through the number of free halogen atoms (fr_halogen) and the logarithm of partition coefficient (MolLogP) as the two most critical molecular descriptors representing the persistence and mobility of substances, respectively. Our screening system exhibits a great generalization capability with area under the receiver operating characteristic curve (AUROC) above 0.951 and is successfully applied to the persistent organic pollutants (POPs), prioritized PMT/vPvM substances, and pesticides. The screening system constructed in this study can serve as an efficient and reliable tool for high-throughput risk assessment and the prioritization of managing emerging contaminants.

show abstract

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

Cited by 30 publications

References 18 publications

Baseline scenarios of heat-related ambulance transportations under climate change in Tokyo, Japan

Baseline scenarios of heat-related ambulance transportations under climate change in Tokyo, Japan

Comparative evaluation of accuracy of the multiple methods of heatstroke risk model including time-stratified case-crossover analysis

Machine Learning-Based Models with High Accuracy and Broad Applicability Domains for Screening PMT/vPvM Substances

Contact Info

Product

Resources

About