REFORMS: Consensus-based Recommendations for Machine-learning-based Science

Science Advances, Journal Year: 2024, Volume and Issue: 10(18)

Published: May 1, 2024

Machine learning (ML) methods are proliferating in scientific research. However, the adoption of these has been accompanied by failures validity, reproducibility, and generalizability. These can hinder progress, lead to false consensus around invalid claims, undermine credibility ML-based science. ML often applied fail similar ways across disciplines. Motivated this observation, our goal is provide clear recommendations for conducting reporting Drawing from an extensive review past literature, we present REFORMS checklist (recommendations machine-learning-based science). It consists 32 questions a paired set guidelines. was developed on basis 19 researchers computer science, data mathematics, social sciences, biomedical sciences. serve as resource when designing implementing study, referees reviewing papers, journals enforcing standards transparency reproducibility.

Language: Английский

---

Causal-Inference Machine Learning Reveals the Drivers of China’s 2022 Ozone Rebound

Environmental Science and Ecotechnology, Journal Year: 2025, Volume and Issue: 24, P. 100524 - 100524

Published: Jan. 11, 2025

Ground-level ozone concentrations rebounded significantly across China in 2022, challenging air quality management and public health. Identifying the drivers of this rebound is crucial for designing effective mitigation strategies. Commonly used methods, such as chemical transport models machine learning, provide valuable insights but face limitations-chemical are computationally intensive, while learning often fails to address confounding factors or establish causality. Here we show that elevated temperatures increased solar radiation, primary meteorological drivers, collectively account 57 % total increase, based on an integrated analysis ground-based monitoring data, satellite observations, reanalysis information using explainable causal inference techniques. Compared year 2021, 90 stations reported increase Formaldehyde Nitrogen ratio, implying a growing sensitivity formation nitrogen oxide levels. These findings highlight significant role changes rebound, urging adoption targeted strategies under climate warming, particularly through varied regional consider existing anthropogenic emission levels prospective biogenic volatile organic compounds. This identification relationships pollution dynamics can support data-driven accurate decision-making.

Language: Английский

---

Integrating artificial intelligence with mechanistic epidemiological modeling: a scoping review of opportunities and challenges

Nature Communications, Journal Year: 2025, Volume and Issue: 16(1)

Published: Jan. 10, 2025

Integrating prior epidemiological knowledge embedded within mechanistic models with the data-mining capabilities of artificial intelligence (AI) offers transformative potential for modeling. While fusion AI and traditional approaches is rapidly advancing, efforts remain fragmented. This scoping review provides a comprehensive overview emerging integrated applied across spectrum infectious diseases. Through systematic search strategies, we identified 245 eligible studies from 15,460 records. Our highlights practical value models, including advances in disease forecasting, model parameterization, calibration. However, key research gaps remain. These include need better incorporation realistic decision-making considerations, expanded exploration diverse datasets, further investigation into biological socio-behavioral mechanisms. Addressing these will unlock synergistic modeling to enhance understanding dynamics support more effective public health planning response. Artificial has improve diseases by incorporating data sources complex interactions. Here, authors conduct use summarise methodological advancements identify gaps.

Language: Английский

---

Characterizing soil Cops Eco-risk in China

Journal of Hazardous Materials, Journal Year: 2025, Volume and Issue: 489, P. 137588 - 137588

Published: Feb. 11, 2025

Language: Английский

---

Healthy aging meta-analyses and scoping review of risk factors across Latin America reveal large heterogeneity and weak predictive models

Nature Aging, Journal Year: 2024, Volume and Issue: 4(8), P. 1153 - 1165

Published: June 17, 2024

Abstract Models of healthy aging are typically based on the United States and Europe may not apply to diverse heterogeneous populations. In this study, our objectives were conduct a meta-analysis assess risk factors cognition functional ability across populations in Latin America scoping review focusing methodological procedures. Our study design included randomized controlled trials cohort, case–control cross-sectional studies using multiple databases, including MEDLINE, Virtual Health Library Web Science. From an initial pool 455 studies, 38 final (28 assessing 10 ability, n = 146,000 participants). results revealed significant but effects for (odds ratio (OR) 1.20, P 0.03, confidence interval (CI) (1.0127, 1.42); heterogeneity: I 2 92.1%, CI (89.8%, 94%)) (OR 0.01, (1.04, 1.39); 93.1%, (89.3%, 95.5%)). Specific had limited effects, especially with moderate impacts demographics mental health marginal status social determinants health. Methodological issues, such as outliers, inter-country differences publication bias, influenced results. Overall, we highlight specific profile associated America. The heterogeneity approaches studying call greater harmonization further regional research understand

Language: Английский

---

Integrating Artificial Intelligence into Causal Research in Epidemiology

Current Epidemiology Reports, Journal Year: 2025, Volume and Issue: 12(1)

Published: March 24, 2025

Language: Английский

---

Short-Term Wind Power Prediction Based on Multi-Parameters Similarity Wind Process Matching and Weighed-Voting-Based Deep Learning Model Selection

IEEE Transactions on Power Systems, Journal Year: 2023, Volume and Issue: 39(1), P. 2129 - 2142

Published: March 15, 2023

Sufficiently accurate short-term wind power prediction is important for the grid dispatch of system. To improve accuracy by selecting suitable model each piece processes, this paper presents a method based on multi-parameters similarity process matching and weighed-voting-based deep learning selection. First, novel presented to match forecast target sample with groups highly similar historical in which 96h-time-scale divided into multiple processes tumbling time window, combinational algorithm that consider four indexes proposed judge among processes. Second, selection method, matched are introduced vote optimal candidate model, select from LSTM, BLSTM, CNN, CNN-LSTM, CNN-BLSTM, SDAE process. Case studies verify effectiveness superiority method. Based new 24h-day-ahead 96h-short-term RMSE can be decreased 0.69% 1.7% 1.15% 2.2% respectively compared single demonstrates

Language: Английский

---

Critical review on data-driven approaches for learning from accidents: Comparative analysis and future research

Safety Science, Journal Year: 2023, Volume and Issue: 171, P. 106381 - 106381

Published: Nov. 27, 2023

Language: Английский

---

Addressing the Challenge of Biomedical Data Inequality: An Artificial Intelligence Perspective

Annual Review of Biomedical Data Science, Journal Year: 2023, Volume and Issue: 6(1), P. 153 - 171

Published: April 27, 2023

Artificial intelligence (AI) and other data-driven technologies hold great promise to transform healthcare confer the predictive power essential precision medicine. However, existing biomedical data, which are a vital resource foundation for developing medical AI models, do not reflect diversity of human population. The low representation in data has become significant health risk non-European populations, growing application opens new pathway this manifest amplify. Here we review current status inequality present conceptual framework understanding its impacts on machine learning. We also discuss recent advances algorithmic interventions mitigating disparities arising from inequality. Finally, briefly newly identified disparity quality among ethnic groups potential

Language: Английский

---

The Commons

Annual Review of Environment and Resources, Journal Year: 2023, Volume and Issue: 48(1), P. 531 - 558

Published: Sept. 14, 2023

Commons—resources used or governed by groups of heterogeneous users through agreed-upon institutional arrangements—are the subject one more successful research programs in social-environmental sciences. This review assesses on commons to accomplish three tasks. First, it surveys theoretical, substantive, and methods-focused achievements field, illustrating how has also influenced natural resource policy making. Second, examines changing trajectories research, emphasizing growing interest researchers new methods application insights social contexts. Third, suggests that can find continuing relevance addressing contemporary future challenges. It highlights directions particular: ( a) strengthening focus issues power equity, b) applying about effective governance collaborative attempts craft societal spaces, c) advancing an emerging emphasis causal analysis taking advantage novel streams large-scale public datasets.

Language: Английский

---