Risk Prediction System Research Articles

Impact of applying machine learning to the electronic medical record on prediction of cancer-associated thrombosis.

409 Background: Cancer-associated thrombosis (CAT) is preventable among high-risk individuals with prophylactic anticoagulation. Risk assessment tools focus only on the initial period after cancer diagnosis; however, some patients may face different risks during different periods of their cancer journey. We developed a longitudinal machine learning system to predict the risk of CAT throughout cancer treatment. Methods: Using electronic health record data at Princess Margaret Cancer Centre, we assembled a cohort of people with thoracic and gastrointestinal cancer receiving systemic treatments between August 1, 2017 and December 31, 2019. We manually labelled 5,000 CT scans and 500 Doppler ultrasounds for the occurrence of CAT. We fine-tuned open-source large language models (LLMs) on those labelled reports, which we then used to automatically detect CAT among all 37,000 CT scans and Dopplers. Finally, we trained longitudinal machine learning systems to predict CAT within 90 days after each cancer treatment. LLMs and the CAT risk prediction system were evaluated among the held-out test cohort of people whose first treatment occurred in 2019. Results: The overall cohort included 2,031 patients and 21,375 treatment sessions. When classifying radiology reports, the fine-tuned LLM achieved an area under the receiver operating characteristic curve (AUROC) of 0.758 for DVT and 0.988 for PE in the test cohort. CAT occurred within 90 days after 6.23% of treatment sessions. The best ML system predicted the risk of CAT within 90 days with an AUROC of 0.651 (95% CI, 0.620-0.678). Considering only the first treatment per patient, treatments within the first 90 days after the first treatment, and treatments after 90 days, the system achieved AUROC of 0.639 (95% CI, 0.568-0.704), 0.599 (95% CI, 0.559-0.642), and 0.741 (95% CI, 0.699-0.783), respectively. Conclusions: Machine learning can longitudinally predict CAT among patients receiving systemic therapy for aerodigestive cancers. These systems could personalize prophylactic anticoagulation by identifying specific periods when patients face sufficient risk to warrant prevention.

Driving risk prediction of urban arterial and collector roads using multi-dimensional real-time data

With the objective of predicting driving risks on urban arterial and collector roads, this paper primarily collects and processes vehicle trajectory data and driver-vehicle-road-environment data. Using Entropy-Technique for Order Preference by Similarity to Ideal Solution (TOPSIS) and K-means, risk scores and levels for the target roads are determined. An original dataset comprising 10,320 samples is established, with risk levels as labels and multi-dimensional data as features. SHapley Additive exPlanations (SHAP) analysis is utilized to extract important features. The performance of four ensemble models are compared, including Random Forest (RF), eXtreme Gradient Boosting (XGBoost), Light Gradient Boosting Machine (LightGBM) and Stacking. Three data resampling techniques are integrated to address data imbalance. The combination of LightGBM and Deep Convolutional Generative Adversarial Network (DCGAN) outperforms other approaches, showing potential for real-time risk prediction. Besides, DCGAN is employed to update the multi-dimensional features. The models' predictive performance is evaluated under various update ratios using F1-score and training time, with LightGBM demonstrating superior accuracy and efficiency compared to its counterparts (RF, XGBoost, and Stacking). This study develops a driving risk prediction system based on multi-dimensional features, offering a theoretical foundation and technical approach to enhance urban road traffic safety management.

Visualization obesity risk prediction system based on machine learning

Obesity is closely associated with various chronic diseases.Therefore, accurate, reliable and cost-effective methods for preventing its occurrence and progression are required. In this study, we developed a visualized obesity risk prediction system based on machine learning techniques, aiming to achieve personalized comprehensive health management for obesity. The system utilized a dataset consisting of 1678 anonymized health examination records, including individual lifestyle factors, body composition, blood routine, and biochemical tests. Ten multi-classification machine learning models, including Random Forest and XGBoost, were constructed to identify non-obese individuals (BMI < 25), class 1 obese individuals (25 ≤ BMI < 30), and class 2 obese individuals (30 ≤ BMI). By evaluating the performance of each model on the test set, we selected XGBoost as the best model and built the visualized obesity risk prediction system based on it. The system exhibited good predictive performance and interpretability, directly providing users with their obesity risk levels and determining corresponding intervention priorities. In conclusion, the developed obesity risk prediction system possesses high accuracy and interactivity, aiding physicians in formulating personalized health management plans and achieving comprehensive and accurate obesity management.

Research on pollution identification and safety thresholds based on the olfactory effect on a halogenated hydrocarbon contaminated site

To effectively address odor control issues at sites contaminated with halogenated hydrocarbons, it is essential to establish an odor risk prediction system for evaluating potential risks that may impact future planning. This research focuses on a representative halogenated hydrocarbon-contaminated site, examining the spatial and temporal distribution characteristics of key pollutants in soil gas. By analyzing odor contribution rates, the study identifies significant odorants in soil gas, which enables the derivation of both probabilistic and deterministic safety thresholds for soil and groundwater based on olfactory effects. The findings indicate that 1,1-dichloroethylene, vinyl chloride, chloroform, and 1,1-dichloroethane are prevalent throughout the contaminated site, displaying elevated concentration levels and substantially influencing the overall contamination extent. These substances are highlighted as critical pollutants requiring attention. Correlation analysis (P < 0.05) reveals a strong relationship between the concentrations of vinyl chloride, 1,1-dichloroethane, and chloroform with groundwater depth and air temperature. Additionally, the analysis of odor activity values (OAV) identified 1,1-dichloroethene, 1,4-dichlorobenzene, chlorobenzene, chloroform, and vinyl chloride as key olfactory factors at the site. The corresponding probabilistic safety thresholds are 0.68, 1.65, 0.50, 7.87, and 3.72 mg kg−1 for soil, and 9.29, 3.46, and 1.09, 69.55, and 47.01 mg L−1 for groundwater, respectively. Among them, the odor risks of chlorobenzene and 1,1-dichloroethylene warrant more attention than soil contamination risks; regarding 1,4-dichlorobenzene, it is recommended to concurrently consider odor risks during human health risk assessment; as for vinyl chloride and chloroform, their odor risks can be largely eliminated based on human health-oriented pollution management.

Prediction of bank credit worthiness through credit risk analysis: an explainable machine learning study

AbstractThe control of credit risk is an important topic in the development of supply chain finance. Financial service providers should distinguish between low- and high-quality customers to predict credit risk accurately. Proper management of credit risk exposure contributes to the long-term viability and profitability of banks, systemic stability, and efficient capital allocation in the economy. Moreover, it benefits the development of supply chain finance. Supply chain finance offers convenient loan transactions that benefit all participants, including the buyer, supplier, and bank. However, poor credit risk management in supply chain finance may cause losses for finance providers and hamper the development of supply chain finance. Machine learning algorithms have significantly improved the accuracy of credit risk prediction systems in supply chain finance. However, their lack of interpretability or transparency makes decision-makers skeptical. Therefore, this study aims to improve AI transparency by ranking the importance of features influencing the decisions made by the system. This study identifies two effective algorithms, Random Forest and Gradient Boosting models, for credit risk detection. The factors that influenced the decision of the models to make them transparent are explicitly illustrated. This study also contributes to the literature on explainable credit risk detection for supply chain finance and provides practical implications for financial institutions to inform decision making.

Triglyceride-glucose index improves risk prediction beyond traditional risk factors and hypertension mediated organ damage in healthy adults

Background and aimsTriglyceride-glucose (TyG) index, a surrogate measure of insulin resistance, is associated with hypertension mediated organ damage (HMOD) and cardiovascular disease. This study investigated the association between TyG index and major adverse cardiovascular events (MACE) and its interaction with traditional risk factors and HMOD. Methods and resultsHealthy subjects recruited from the general population were thoroughly examined and followed for MACE using nation-wide registries. Cox proportional hazard models were used to calculate the association between TyG index and MACE occurrence. Models were adjusted for Systematic Coronary Risk Evaluation (SCORE) risk factors, pulse wave velocity, left ventricular mass index, carotid atherosclerotic plaque status, and microalbuminuria. Continuous net reclassification and Harrell's Concordance index (C-index) were used to assess the added prognostic value of TyG index.During a follow-up period of mean 15.4 ± 4.7 years, MACE were observed in 332 (17%) of 1970 included participants. TyG index was associated with MACE; HR = 1.44 [95%CI:1.30–1.59] per standard deviation. After adjustment for traditional cardiovascular (CV) risk factors, HR was 1.16 [95%CI:1.03–1.31]. The association between TyG index and MACE remained significant after further adjustment for each HMOD component. However, this finding was evident only in subjects aged 41 or 51 years (HR = 1.39; 95%CI:1.15–1.69). Including TyG index in a risk model based on traditional CV risk factors improved C-index with 0.005 (P = 0.042). ConclusionIn this population-based study of healthy middle-aged subjects, TyG index was associated with MACE independently of traditional CV risk factors and HMOD. TyG index may have a potential role in future risk prediction systems.

Promising algorithms to perilous applications: a systematic review of risk stratification tools for predicting healthcare utilisation

ObjectivesRisk stratification tools that predict healthcare utilisation are extensively integrated into primary care systems worldwide, forming a key component of anticipatory care pathways, where high-risk individuals are targeted by preventative...

Assessing eco-geographic influences on COVID-19 transmission: a global analysis.

COVID-19 has been massively transmitted for almost 3years, and its multiple variants have caused serious health problems and an economic crisis. Our goal was to identify the influencing factors that reduce the threshold of disease transmission and to analyze the epidemiological patterns of COVID-19. This study served as an early assessment of the epidemiological characteristics of COVID-19 using the MaxEnt species distribution algorithm using the maximum entropy model. The transmission of COVID-19 was evaluated based on human factors and environmental variables, including climate, terrain and vegetation, along with COVID-19 daily confirmed case location data. The results of the SDM model indicate that population density was the major factor influencing the spread of COVID-19. Altitude, land cover and climatic factor showed low impact. We identified a set of practical, high-resolution, multi-factor-based maximum entropy ecological niche risk prediction systems to assess the transmission risk of the COVID-19 epidemic globally. This study provided a comprehensive analysis of various factors influencing the transmission of COVID-19, incorporating both human and environmental variables. These findings emphasize the role of different types of influencing variables in disease transmission, which could have implications for global health regulations and preparedness strategies for future outbreaks.

A Multi-Satellite Space Environment Risk Prediction and Real-Time Warning System for Satellite Safety Management

In response to the need for a space security situation assessment during orbit, the multi-satellite space environmental risk prediction and early warning system is based on the detection results of the space weather payload of the Fengyun 4A and 4B satellites, as well as the prediction results of the National Space Weather Center, for the first time. By comprehensively utilizing some open-source data, it is the first time that we have achieved a 24 h advanced prediction of the space environment high-energy proton, low-energy particle, and high-energy electron risks for the safety of the Fengyun-series high-orbit satellites, and a real-time warning of satellite single-event upset, surface charging, and deep charging risks. The automation system has preliminarily achieved an intelligent space risk assessment for the safety of multiple stationary meteorological satellites, effectively improving the application efficiency of the space environmental data and the products of Fengyun-series satellites. The business status is stable in operation, and the resulting error between the predicted results of various risk indices and the measured data was less than one level. The warning accuracy was better than 90%. This article uses the system for the first time, to use Fengyun satellite data to, accurately and in a timely manner, predict and warn us about the low-energy particles and surface charging high-risk levels of the Fengyun 4A and 4B satellites during the typical space weather event on 21 April 2023, in response to the impact of complex spatial environmental factors on the safety of Fengyun-series high-orbit satellites. The construction and operation of a multi-satellite space environmental risk prediction and early warning system can provide a reference for the safety work of subsequent satellite ground system operations.

Noninvasive Risk Prediction Models for Heart Failure Using Proportional Jaccard Indices and Comorbidity Patterns.

In the post-coronavirus disease 2019 (COVID-19) era, remote diagnosis and precision preventive medicine have emerged as pivotal clinical medicine applications. This study aims to develop a digital health-monitoring tool that utilizes electronic medical records (EMRs) as the foundation for performing a non-random correlation analysis among different comorbidity patterns for heart failure (HF). Novel similarity indices, including proportional Jaccard index (PJI), multiplication of the odds ratio proportional Jaccard index (OPJI), and alpha proportional Jaccard index (APJI), provide a fundamental framework for constructing machine learning models to predict the risk conditions associated with HF. Our models were constructed for different age groups and sexes and yielded accurate predictions of high-risk HF across demographics. The results indicated that the optimal prediction model achieved a notable accuracy of 82.1% and an area under the curve (AUC) of 0.878. Our noninvasive HF risk prediction system is based on historical EMRs and provides a practical approach. The proposed indices provided simple and straightforward comparative indicators of comorbidity pattern matching within individual EMRs. All source codes developed for our noninvasive prediction models can be retrieved from GitHub.

Prediction of viral spillover risk based on the mass action principle

Infectious zoonotic disease emergence, through spillover events, is of global concern and has the potential to cause significant harm to society, as recently demonstrated by COVID-19. More than 70% of the 400 infectious diseases that emerged in the past five decades have a zoonotic origin, including all recent pandemics. There have been several approaches used to predict the risk of spillover through some of the known or suspected infectious disease emergence drivers, largely using correlative approaches. Here, we predict the spatial distribution of spillover risk by approximating general transmission through animal and human interactions. These mass action interactions are approximated through the multiplication of the spatial distribution of zoonotic virus diversity and human population density. Although our results indicate higher risk in regions along the equator and in Southeast Asia where both virus diversity and human population density are high, it should be noted that this is primarily a conceptual exercise. We compared our spillover risk map to key factors, including the model inputs of zoonotic virus diversity estimate map, human population density map, and the spatial distribution of species richness. Despite the limitations of this approach, this viral spillover map is a step towards developing a more comprehensive spillover risk prediction system to inform global monitoring.

Investigating the Impact of AI on Shared Decision-Making in Post-Kidney Transplant Care (PRIMA-AI): Protocol for a Randomized Controlled Trial.

Patients after kidney transplantation eventually face the risk of graft loss with the concomitant need for dialysis or retransplantation. Choosing the right kidney replacement therapy after graft loss is an important preference-sensitive decision for kidney transplant recipients. However, the rate of conversations about treatment options after kidney graft loss has been shown to be as low as 13% in previous studies. It is unknown whether the implementation of artificial intelligence (AI)-based risk prediction models can increase the number of conversations about treatment options after graft loss and how this might influence the associated shared decision-making (SDM). This study aims to explore the impact of AI-based risk prediction for the risk of graft loss on the frequency of conversations about the treatment options after graft loss, as well as the associated SDM process. This is a 2-year, prospective, randomized, 2-armed, parallel-group, single-center trial in a German kidney transplant center. All patients will receive the same routine post-kidney transplant care that usually includes follow-up visits every 3 months at the kidney transplant center. For patients in the intervention arm, physicians will be assisted by a validated and previously published AI-based risk prediction system that estimates the risk for graft loss in the next year, starting from 3 months after randomization until 24 months after randomization. The study population will consist of 122 kidney transplant recipients >12 months after transplantation, who are at least 18 years of age, are able to communicate in German, and have an estimated glomerular filtration rate <30 mL/min/1.73 m2. Patients with multi-organ transplantation, or who are not able to communicate in German, as well as underage patients, cannot participate. For the primary end point, the proportion of patients who have had a conversation about their treatment options after graft loss is compared at 12 months after randomization. Additionally, 2 different assessment tools for SDM, the CollaboRATE mean score and the Control Preference Scale, are compared between the 2 groups at 12 months and 24 months after randomization. Furthermore, recordings of patient-physician conversations, as well as semistructured interviews with patients, support persons, and physicians, are performed to support the quantitative results. The enrollment for the study is ongoing. The first results are expected to be submitted for publication in 2025. This is the first study to examine the influence of AI-based risk prediction on physician-patient interaction in the context of kidney transplantation. We use a mixed methods approach by combining a randomized design with a simple quantitative end point (frequency of conversations), different quantitative measurements for SDM, and several qualitative research methods (eg, records of physician-patient conversations and semistructured interviews) to examine the implementation of AI-based risk prediction in the clinic. ClinicalTrials.gov NCT06056518; https://clinicaltrials.gov/study/NCT06056518. PRR1-10.2196/54857.

The application of mammography imaging in the diagnosis and prediction of breast diseases

Breast diseases pose a significant threat to women’s health, so early detection and treatment are extremely important. In this context, early disease identification has become crucial in the diagnosis and treatment of breast diseases. This paper begins by outlining the pivotal role of mammography in the early diagnosis of breast cancer, comparing the structural similarities and differences between normal and diseased breast tissues. This comparison underscores the primary role of mammography in the diagnosis and treatment of breast diseases. Additionally, our paper provides an overview of fundamental concepts related to breast cancer detection, diagnosis, and prediction systems. It delves into the latest research developments in auxiliary diagnostic detection, examination, and risk prediction systems associated with breast cancer. Our objective is to offer a comprehensive understanding of the role of computer-aided detection, diagnosis, and prediction systems in breast diseases, fostering further development and application. This work aims to explore and drive innovation in the field, enhance early detection rates of breast diseases, and guide readers towards novel directions, thus contributing to female healthcare management.

Modelling the dynamic basic reproduction number of dengue based on MOI of Aedes albopictus derived from a multi-site field investigation in Guangzhou, a subtropical region

BackgroundMore than half of the global population lives in areas at risk of dengue (DENV) transmission. Developing an efficient risk prediction system can help curb dengue outbreaks, but multiple variables, including mosquito-based surveillance indicators, still constrain our understanding. Mosquito or oviposition positive index (MOI) has been utilized in field surveillance to monitor the wild population density of Aedes albopictus in Guangzhou since 2005.MethodsBased on the mosquito surveillance data using Mosq-ovitrap collection and human landing collection (HLC) launched at 12 sites in Guangzhou from 2015 to 2017, we established a MOI-based model of the basic dengue reproduction number (R0) using the classical Ross-Macdonald framework combined with a linear mixed-effects model.ResultsDuring the survey period, the mean MOI and adult mosquito density index (ADI) using HLC for Ae. albopictus were 12.96 ± 17.78 and 16.79 ± 55.92, respectively. The R0 estimated from the daily ADI (ADID) showed a significant seasonal variation. A 10-unit increase in MOI was associated with 1.08-fold (95% CI 1.05, 1.11) ADID and an increase of 0.14 (95% CI 0.05, 0.23) in the logarithmic transformation of R0. MOI-based R0 of dengue varied by month and average monthly temperature. During the active period of Ae. albopictus from April to November in Guangzhou region, a high risk of dengue outbreak was predicted by the MOI-based R0 model, especially from August to October, with the predicted R0 > 1. Meanwhile, from December to March, the estimates of MOI-based R0 were < 1.ConclusionsThe present study enriched our knowledge about mosquito-based surveillance indicators and indicated that the MOI of Ae. albopictus could be valuable for application in estimating the R0 of dengue using a statistical model. The MOI-based R0 model prediction of the risk of dengue transmission varied by month and temperature in Guangzhou. Our findings lay a foundation for further development of a complex efficient dengue risk prediction system.Graphical

Clinical and imaging predictors for hemorrhagic transformation of acute ischemic stroke after endovascular thrombectomy.

Hemorrhagic transformation (HT) is a common complication of endovascular thrombectomy (EVT) in patients with acute ischemic stroke (AIS). Our study aims to investigate the clinical and imaging predictors of HT and symptomatic intracranial hemorrhage (sICH) in patients who underwent EVT. A retrospective analysis of 118 patients undergoing EVT for acute anterior circulation stroke was performed. Potential clinical and imaging predictors of all patients were collected and multivariate logistic regression was performed. The risk prediction system was constructed according to the multivariate logistic regression results. The incidence of HT and sICH after EVT were 46.6% and 15.3%, respectively. The multivariate logistic regression results showed that Alberta Stroke Program Early CT Score (ASPECTS) (p = .001, odds ratio [OR] = 0.367, 95% [confidence interval] CI, 0.201-0.670), collateral status (p<.001, OR = 0.117, 95% CI, 0.042-0.325), relative cerebral blood flow (CBF) ratio (p = .025, OR = 0.943, 95% CI, 0.895-0.993), and blood glucose on admission (p = .012, OR = 1.258, 95% CI, 1.053-1.504) were associated with HT. While for sICH, collateral circulation (p = .007, OR = 0.148, 95% CI, 0.037-0.589), ASPECTS (p = .033, OR = 0.510, 95% CI, 0.274-0.946), and blood glucose (p = .005, OR = 1.304, 95% CI, 1.082-1.573) were independent factors. The predictive model for HT after EVT was established, and the sensitivity and specificity of it were 90.9% and 79.4%, respectively, with the area under the curve of 90.0% (84.5%-95.4%). Collateral status, ASPECTS, relative CBF ratio, and blood glucose on admission were predictors for HT in AIS patients, while collateral status, ASPECTS, and blood glucose on admission were also predictors for sICH. In addition, the established predictive model showed good diagnostic value for prediction of HT after EVT.

Intersection collision risk evaluation and active collision avoidance strategies for autonomous vehicles

In order to ensure that the autonomous vehicle can predict and taking actions to avoid the collision in time when facing the obstacles with intersection collision risk, an intersection collision risk prediction system is proposed in this paper, and two kinds of active obstacle avoidance strategies are designed according to the system: braking strategy and steering strategy. The position information of the obstacle is predicted by Fractional extended Kalman filter, the collision risk rate is determined by the time difference between the vehicle and the obstacle through the intersection point, and a neural network is trained to quickly give the collision risk of the vehicle and the obstacle. Braking strategy and steering strategy are formulated according to collision risk, the braking deceleration and Sigmoid path parameters are given. Finally, the simulation results of PreScan and MATLAB show that the collision risk prediction system can accurately predict the collision between vehicles and obstacles, the braking and steering strategies can effectively avoid the collision.

PENILAIAN KREDIT UNTUK UKM MENGGUNAKAN HYBRID BWM DAN TOPSIS PADA BANK SYARIAH DI INDONESIA

This research develops and validates a SME credit risk prediction system by applying a multi-criteria credit scoring model. The model was built using the best-worst method (BWM) and the preferred order of similarity to ideal solution (TOPSIS) technique. BWM first sets the weighting criteria and TOPSIS is applied to evaluate SMEs. Real-life case studies are examined to demonstrate the effectiveness of the proposed model. The results showed that that collateral can be bound in accordance with the provisions, history of credit, and current ratio were the most important factors in lending, followed by customer business sector category, continuity of supply of raw materials, and collateral adequacy requirements. This model can help financial institutions provide an easy way to identify potential SMEs for loans and encourage further research into alternative approaches.

Integrative approach for modern health risk modeling and predicting in patients through artificial intelligence method

This research suggests a modeling approach for health risk prediction that utilizes an ambient environment and Artificial Intelligence (AI). The proposed AI-based Health Risk Modeling and Predicting System (AI-HRMPS) included gathering medical records from chronic illness patients, including Electronic Health Records (EHR), Personal Health Records (PHR), medical records, and environmental variables from a health portal. Diverse data is combined via choosing, cleaning, modeling, and assessing raw data, followed by data production. Sensor data is standardized by converting the time-domain details to frequency-domain details. The standardized input is processed using an AI to provide an ambient environment. A health risk prediction system has been proposed to analyze specific health issues about environmental factors. The program utilizes ambient context patterns identified via metadata and AI. The risk prediction framework is integrated into a person's risk alert/prevention mechanism. The system might substantially influence healthcare and AI studies, ultimately enhancing the future society's standard of life.

Modeling the Risk of Liver Cancer in Adults: A Machine Learning Investigation into the Role of Obesity and Overweight

Background: Liver cancer is a global health concern, with overweight and obese individuals exhibiting an increased risk of its development. Understanding the interplay between obesity-related factors and liver cancer incidence is crucial for early prediction and intervention. Aim: The aim of this investigation was to construct and validate an extreme gradient boosting (XGBoost) based machine learning model for the purpose of establishing a one-year liver cancer risk prediction system specifically tailored to overweight and obese patients. In addition, this study sought to compare the predictive performance of the XGBoost model with those of a random forest model and a logistic regression model, while also identifying the most influential predictive features for liver cancer incidence. Methods: A comprehensive retrospective analysis was conducted on MIMIC III data comprising 2,354 patients. To predict the risk of liver cancer development, three machine learning models were developed: XGBoost, random forest, and logistic regression. Feature selection was executed using a stepwise regression procedure encompassing both forward selection and backward elimination. Results: The stepwise regression technique unveiled 14 predictive factors for liver cancer incidence. Among the patient cohort, 132 individuals developed liver cancer within a year of follow-up, while 2,222 did not. Notably, most liver cancer cases occurred in male patients (60%). Statistically significant differences were observed between patients with liver cancer and those without, in terms of age, gender, total bilirubin, platelet, albumin, chloride, potassium, sodium, prothrombin time (PT) and alanine aminotransferase (ALT). The XGBoost model exhibited an impressive area under the receiver operating characteristic curve (AUROC) of 99%, Random Forest (RF) of 99%, and Logistic Regression (LR) of 90%. In a multivariate analysis, total bilirubin, creatinine levels, age, gender, ALT, alkaline phosphate (ALP), PT, calcium, and chloride emerged as independent predictors for liver cancer incidence. Conclusion: The XGBoost model demonstrated superior predictive performance when compared to the RF and LR models. If corroborated through prospective studies, the XGBoost model may prove to be a valuable tool for the early prediction of liver cancer risk in overweight or obese individuals. Such predictive capabilities could, in turn, facilitate the implementation of timely preventive interventions against liver cancer.

A machine learning-based universal outbreak risk prediction tool

In order to prevent and control the increasing number of serious epidemics, the ability to predict the risk caused by emerging outbreaks is essential. However, most current risk prediction tools, except EPIRISK, are limited by being designed for targeting only one specific disease and one country. Differences between countries and diseases (e.g., different economic conditions, different modes of transmission, etc.) pose challenges for building models with cross-country and cross-disease prediction capabilities. The limitation of universality affects domestic and international efforts to control and prevent pandemic outbreaks. To address this problem, we used outbreak data from 43 diseases in 206 countries to develop a universal risk prediction system that can be used across countries and diseases. This system used five machine learning models (including Neural Network XGBoost, Logistic Boost, Random Forest and Kernel SVM) to predict and vote together to make ensemble predictions. It can make predictions with around 80%–90 % accuracy from economic, cultural, social, and epidemiological factors. Three different datasets were designed to test the performance of ML models under different realistic situations. This prediction system has strong predictive ability, adaptability, and generality. It can give universal outbreak risk assessment that are not limited by border or disease type, facilitate rapid response to pandemic outbreaks, government decision-making and international cooperation.