Prediction Algorithm Research Articles

Interethnic validation of an artificial intelligence algorithm for prediction of atrial fibrillation using sinus rhythm electrocardiogram

Abstract Background Previous research has demonstrated acceptable diagnostic accuracy of artificial intelligence (AI)-enabled sinus rhythm (SR) electrocardiogram (ECG) interpretation for identifying paroxysmal atrial fibrillation (AF). However, interethnic validations of the AI algorithms have not been widely conducted. Purpose We aimed to develop our own AI model for the identification of paroxysmal AF based on SR ECGs in the Korean population and to validate its diagnostic performance in Brazilian citizens. Methods We trained a Transformer-based vision network on 90% of a dataset comprising 808,194 ECGs from 121,282 patients at Seoul National University Bundang Hospital (2003-2020). The remaining 10% of the dataset was used for internal validation. The model was trained to compute a risk score for paroxysmal AF or new-onset AF within 2 years. External validation was conducted using non-AF ECGs from the CODE 15% dataset provided by the Telehealth Network of Minas Gerais, Brazil. Results In internal validation, our AI model achieved an Area Under the Receiver Operating Characteristic Curve (AUROC) of 0.911 (95% CI: 0.902 - 0.921), with a sensitivity of 78.1% and a specificity of 89.0%. Subgroup analyses showed an AUROC of 0.893 (95% CI: 0.876 - 0.909) for patients in routine health checkups or outpatient settings, and 0.854 for patients with "Normal ECG" interpretations. In external validation with the CODE 15% dataset, the AI model exhibited an AUROC of 0.884 (95% CI: 0.869-0.900), which increased to 0.906 (95% CI: 0.893-0.919) when adjusted for age and sex (Figure 1). In the subset of patients with "Normal ECG" interpretations, the AUROC was 0.826 (95% CI: 0.769-0.883), increasing to 0.861 (95% CI: 0.814-0.908) after applying the same adjustments. Conclusions Our AI-powered SR ECG interpretation model demonstrated excellent diagnostic performance in predicting paroxysmal AF, with valid performance in the Brazilian population as well. This suggests that the model has potential for broad application across different ethnic groups.Figure 1

The added value of deep learning to submaximal exercise electrocardiogram for the 10-year prediction of major adverse cardiovascular and cerebrovascular events

Abstract Background The added value of exercise ECG data to traditional risk factors in predicting cardiovascular events remains unclear. Deep learning of ECG signals has demonstrated state-of-the-art performance in detecting subtle abnormalities indicative of cardiovascular disease. We investigated whether deep learning analysis of exercise ECGs could improve the prediction of major cardiovascular and cerebrovascular events (MACCE) with respect to established risk models in a large population-based cohort. Purpose To assess the added value of submaximal exercise ECG measurements to established cardiovascular risk prediction models. Methods We obtained ECG recordings from 41,076 UK Biobank participants without cardiovascular disease (CVD) who underwent a submaximal exercise ECG test. We obtained ECG recordings from 41,076 UK Biobank participants without cardiovascular disease (CVD) who underwent a submaximal exercise ECG test. We created two deep neural network ECG risk scores: one derived from conventional ECG parameters, measured at rest, peak exercise and late-stage recovery, and another based on the complete ECG (Q-ECG). We assessed the added value of conventional ECG parameters and Q-ECG risk scores to the UK's current prediction algorithm (QRISK3). We estimate the association between ECG parameters and MACCE, using Cox Proportional hazard models, following adjustment for traditional risk factors. All models were internally validated via 5-fold cross-validation and 1000 bootstrap iterations. Predictive performance was evaluated using Harrel's C-index, Net Reclassification Index (NRI) and net benefit. Findings: Incident MACCE was reported in 4,082 (9.9%) individuals in the study population and 3,463(9.7%) individuals with valid ECG parameters over a median follow-up period of 12.5 years. We found combined conventional ECG and Q-ECG scores were independently associated with MACCE, following adjustment for multiple testing: adjusted hazard ratio [HZ] = 1.76 (95% CI:1.63-1.91); HZ = 1.14 (95% CI:1.10-1.18), respectively, per standard deviation increase. Both conventional ECG parameters and Q-ECG were predictive of MACCE, independent of clinical risk factors, C-index = 0.64 (95%CI 0.63-0.65); net benefit = 0.09(95% CI 0.07 - 0.11) and C-index = 0.56 (95% CI 0.55-0.57); net benefit = 0.07(95% CI 0.05 - 0.09). ECG measurement's modestly improved model discrimination over the bassline QRISK3 risk score when combined with QRISK3 risk factors for conventional markers and Q-ECG score, respectively; ΔC-index 0.03 (95% CI: 0.02 – 0.04) and ΔC-index 0.03 (95% CI: 0.02 – 0.03); However, we observed no significant improvements in classification at the current recommended threshold of 10%. Conclusion In individuals without a history of prior cardiovascular disease, ECG measures independently predict the risk of MACCE. When combined with QRISK3, neural-network-derived ECG risk scores marginally improve cardiovascular risk prediction over QRISK3.

A comprehensive comparison of deep learning-based compound-target interaction prediction models to unveil guiding design principles

The evaluation of compound-target interactions (CTIs) is at the heart of drug discovery efforts. Given the substantial time and monetary costs of classical experimental screening, significant efforts have been dedicated to develop deep learning-based models that can accurately predict CTIs. A comprehensive comparison of these models on a large, curated CTI dataset is, however, still lacking. Here, we perform an in-depth comparison of 12 state-of-the-art deep learning architectures that use different protein and compound representations. The models were selected for their reported performance and architectures. To reliably compare model performance, we curated over 300 thousand binding and non-binding CTIs and established several gold-standard datasets of varying size and information. Based on our findings, DeepConv-DTI consistently outperforms other models in CTI prediction performance across the majority of datasets. It achieves an MCC of 0.6 or higher for most of the datasets and is one of the fastest models in training and inference. These results indicate that utilizing convolutional-based windows as in DeepConv-DTI to traverse trainable embeddings is a highly effective approach for capturing informative protein features. We also observed that physicochemical embeddings of targets increased model performance. We therefore modified DeepConv-DTI to include normalized physicochemical properties, which resulted in the overall best performing model Phys-DeepConv-DTI. This work highlights how the systematic evaluation of input features of compounds and targets, as well as their corresponding neural network architectures, can serve as a roadmap for the future development of improved CTI models.Scientific contributionThis work features comprehensive CTI datasets to allow for the objective comparison and benchmarking of CTI prediction algorithms. Based on this dataset, we gained insights into which embeddings of compounds and targets and which deep learning-based algorithms perform best, providing a blueprint for the future development of CTI algorithms. Using the insights gained from this screen, we provide a novel CTI algorithm with state-of-the-art performance.

SCORE2-ASIA risk prediction algorithms: revised models to estimate 10-year risk of cardiovascular disease in the Asia Pacific Region

Abstract Background Most current cardiovascular disease (CVD) prediction tools have been developed and validated solely in Western countries, while there are significant differences between Asian and Western populations in risk factor distributions and CVD incidence patterns. This study outlines the recalibration and validation of the SCORE2-ASIA model to estimate 10-year fatal and non-fatal CVD risk among adults aged 40-69 years with no previous CVD or diabetes in the Asia Pacific Region. Methods The sex-specific and competing risk-adjusted SCORE2 models were systematically recalibrated to reflect CVD incidence in the Asian clinical practice. We defined four risk regions in Asia according to country-specific CVD mortality rates (low risk (<100 CVD deaths per 100,000), moderate risk (100 to <150 CVD deaths per 100,000), high risk (150 to <300 CVD deaths per 100,000), and very high risk (≥300 CVD deaths per 100,000)). Models were recalibrated using region representative data on expected incidences and risk factor distributions, utilising the methodological approaches developed in recalibration of the SCORE2 models for Europe. The SCORE-ASIA model was externally validated in all risk regions using population-based cohort data from independents cohort data. To ensure the European coefficients well applied to Asian populations, the SCORE2 model derivation process was repeated in external validation cohorts. Results Region-specific incidence was estimated using country-specific data on CVD mortality and data on the ratio of fatal to total events on 9,496,212 individuals from 5 Asian countries (496,212 CVD events). For external validation, we analysed data from 12 additional cohorts in 13 Asian countries (4,775,976 individuals, 176,011 CVD events). After applying the derived risk prediction models to external validation cohorts, the pooled external validation C-index was 0.709 (95%CI 0.676-0.743). In the nationally representative CDMD cohort from Singapore (population completely separate from recalibration data) there was adequate agreement between predicted and observed risks. Risk factor coefficients derived in Asian populations were largely similar to the original SCORE2 coefficients, indicating that the original SCORE2 coefficients apply well to the Asian population. Predicted CVD risk varied several-fold across Asian risk regions. For example, the estimated 10-year CVD risk for a 50-year-old non-smoker, with a systolic blood pressure of 140mmHg, total cholesterol of 5.5mmol/L, and HDL-cholesterol of 1.3mmol/L, ranged from 12% for men in low-risk countries to 26% for men in very-high-risk countries, and from 7% for women in low-risk countries to 26% for women in very-high-risk countries. Conclusions The SCORE2-ASIA model accurately predicts 10-year risk of CVD for apparently healthy people in Asian countries after systematic recalibration, thereby enhancing the identification of individuals at higher risk of developing CVD across the Asia Pacific Region.

The T cell receptor β chain repertoire of tumor infiltrating lymphocytes improves neoantigen prediction and prioritization.

In the realm of cancer immunotherapy, the meticulous selection of neoantigens plays a fundamental role in enhancing personalized treatments. Traditionally, this selection process has heavily relied on predicting the binding of peptides to human leukocyte antigens (pHLA). Nevertheless, this approach often overlooks the dynamic interaction between tumor cells and the immune system. In response to this limitation, we have developed an innovative prediction algorithm rooted in machine learning, integrating T cell receptor β chain (TCRβ) profiling data from colorectal cancer (CRC) patients for a more precise neoantigen prioritization. TCRβ sequencing was conducted to profile the TCR repertoire of tumor-infiltrating lymphocytes (TILs) from 28 CRC patients. The data unveiled both intra-tumor and inter-patient heterogeneity in the TCRβ repertoires of CRC patients, likely resulting from the stochastic utilization of V and J segments in response to neoantigens. Our novel combined model integrates pHLA binding information with pHLA-TCR binding to prioritize neoantigens, resulting in heightened specificity and sensitivity compared to models using individual features alone. The efficacy of our proposed model was corroborated through ELISpot assays on long peptides, performed on four CRC patients. These assays demonstrated that neoantigen candidates prioritized by our combined model outperformed predictions made by the established tool NetMHCpan. This comprehensive assessment underscores the significance of integrating pHLA binding with pHLA-TCR binding analysis for more effective immunotherapeutic strategies.

Photovoltaic Maximum Power Point Tracking Technology Based on Power Prediction Algorithm Combined with Variable Step Length Disturbance Observation Method

Under partial shading conditions, the system operating sequence of photovoltaic cells does not fall on a single characteristic curve, and the traditional maximum power point tracking (MPPT) control algorithm is prone to causing the system output power to oscillate around the maximum power point. In the case of multiple peaks, the traditional MPPT algorithm is prone to being trapped in a local optimum solution. This paper proposes an MPPT control algorithm based on the WOA-RF prediction algorithm combined with a variable step disturbance observation method. By establishing a photovoltaic system simulation model, the improved algorithm is verified through simulation. The improved MPPT control algorithm can adaptively adjust the step size under different conditions to ensure that the system can quickly and accurately track the maximum power point. At the same time, by optimizing the algorithm logic and parameter settings, it can effectively avoid problems such as local optimum solutions and instability in the system, and improve the system convergence speed and tracking accuracy.

Artificial intelligence in cardiology: a machine learning model for supraventricular tachycardia discrimination

Abstract Background Distinguishing atrioventricular nodal reentrant tachycardia (AVNRT) from orthodromic atrioventricular reentrant tachycardia (AVRT) presents a diagnostic challenge, with clinical and electrocardiogram (ECG) features playing a crucial role in differential diagnosis. Purpose The aim of this study was to devise and validate a predictive algorithm that utilises basic clinical and ECG parameters to accurately classify AVNRT versus AVRT. Furthermore, this algorithm was integrated into a user-friendly, web-based tool to support clinical decision-making. Methods We conducted a prospective analysis of 704 patients (median age: 53.6 years; 59.7% female) with electrophysiologically confirmed AVNRT (n=535) or AVRT (n=169). A comprehensive evaluation was performed using a structured questionnaire tailored to arrhythmia-specific symptoms. The study utilised advanced machine learning techniques for classification, employing feature selection methods to enhance predictive accuracy. The analysis included 19 demographic, clinical, and ECG variables, adopting an 80-20% split for training and internal validation. The algorithm's performance was evaluated against a model comprising all variables (Full Model), expert ECG interpretations (ExpertECG), objective ECG criteria (ECG model), and a baseline dummy classifier, with efficacy assessed by the Area Under the Receiver Operating Characteristic curve (AUC) and the Area Under the Precision-Recall Curve (AUPRC). Results The optimised model, utilising the XGBoost algorithm with six critical variables (initial age at first episode, biological sex, presence of neck palpitations, pseudoR wave in lead V1/pseudoS wave in inferior leads, visibility of retrograde P-waves, and tachycardia cycle length), exhibited superior performance, achieving an AUC of 0.916 and an AUPRC of 0.823. This model significantly outperformed other comparative models in terms of predictive accuracy (Figure 1). Implementation of the model into a web-based interface was accomplished using Shiny, enhancing its accessibility for clinical application (Figure 2). Conclusions This study illustrates the efficacy of employing machine learning algorithms to utilise bedside clinical and ECG variables for accurately differentiating between AVNRT and AVRT. The creation of a web-based calculator marks a significant advancement, empowering clinicians, especially non-specialists, with a potent tool for providing personalised diagnostic insights. Future research should focus on external validation to ascertain the model's generalisability and its applicability in broader clinical settings.Figure 1Figure 2

The T cell receptor β chain repertoire of tumor infiltrating lymphocytes improves neoantigen prediction and prioritization

In the realm of cancer immunotherapy, the meticulous selection of neoantigens plays a fundamental role in enhancing personalized treatments. Traditionally, this selection process has heavily relied on predicting the binding of peptides to human leukocyte antigens (pHLA). Nevertheless, this approach often overlooks the dynamic interaction between tumor cells and the immune system. In response to this limitation, we have developed an innovative prediction algorithm rooted in machine learning, integrating T cell receptor β chain (TCRβ) profiling data from colorectal cancer (CRC) patients for a more precise neoantigen prioritization. TCRβ sequencing was conducted to profile the TCR repertoire of tumor-infiltrating lymphocytes (TILs) from 28 CRC patients. The data unveiled both intra-tumor and inter-patient heterogeneity in the TCRβ repertoires of CRC patients, likely resulting from the stochastic utilization of V and J segments in response to neoantigens. Our novel combined model integrates pHLA binding information with pHLA-TCR binding to prioritize neoantigens, resulting in heightened specificity and sensitivity compared to models using individual features alone. The efficacy of our proposed model was corroborated through ELISpot assays on long peptides, performed on four CRC patients. These assays demonstrated that neoantigen candidates prioritized by our combined model outperformed predictions made by the established tool NetMHCpan. This comprehensive assessment underscores the significance of integrating pHLA binding with pHLA-TCR binding analysis for more effective immunotherapeutic strategies.

Predicting the Wear Amount of Tire Tread Using 1D-CNN.

Since excessively worn tires pose a significant risk to vehicle safety, it is crucial to monitor tire wear regularly. This study aimed to verify the efficient tire wear prediction algorithm proposed in a previous modeling study, which minimizes the required input data, and use driving test data to validate the method. First, driving tests were conducted with tires at various wear levels to measure internal accelerations. The acceleration signals were then screened using empirical functions to exclude atypical data before proceeding with the machine learning process. Finally, a tire wear prediction algorithm based on a 1D-CNN with bottleneck features was developed and evaluated. The developed algorithm showed an RMSE of 5.2% (or 0.42 mm) using only the acceleration signals. When tire pressure and vertical load were included, the prediction error was reduced by 11.5%, resulting in an RMSE of 4.6%. These findings suggest that the 1D-CNN approach is an efficient method for predicting tire wear states, requiring minimal input data. Additionally, it supports the potential usefulness of the intelligent tire technology framework proposed in the modeling study.

Anchoring temporal convolutional networks for epileptic seizure prediction.

Accurate and timely prediction of epileptic seizures is crucial for empowering patients to mitigate their impact or prevent them altogether. Current studies predominantly focus on short-term seizure predictions, which causes the prediction time to be shorter than the onset of antiepileptic, thus failing to prevent seizures. However, longer epilepsy prediction faces the problem that as the preictal period lengthens, it increasingly resembles the interictal period, complicating differentiation. To address these issues, we employ the sample entropy method for feature extraction from electroencephalography (EEG) signals. Subsequently, we introduce the Anchoring Temporal Convolutional Networks (ATCN) model for longer-term, patient-specific epilepsy prediction. ATCN utilizes dilated causal convolutional networks to learn time-dependent features from previous data, capturing temporal causal correlations within and between samples. Additionally, the model also incorporates anchoring data to enhance the performance of epilepsy prediction further. Finally, we proposed a multilayer sliding window prediction algorithm for seizure alarms. Evaluation on the Freiburg intracranial EEG dataset shows our approach achieves 100% sensitivity, a false prediction rate (FPR) of 0.08 per hour and an average prediction time (APT) of 99.98 minutes. Using the CHB-MIT scalp EEG dataset, we a achieve 97.44% sensitivity, an FPR of 0.11 per hour, and an APT of 92.99 minutes. These results demonstrate that our approach is adequate for seizure prediction over a more extended prediction range on intracranial and scalp EEG datasets. The average prediction time of our approach exceeds the typical onset time of antiepileptic. This approach is particularly beneficial for patients who need to take medication at regular intervals, as they may only need to take their medication when our method issues an alarm. This capability has the potential to prevent seizures, which will greatly improving patients' quality of life.

Coronary inflammation stratifies risk in patients undergoing coronary CT angiography: the Oxford Risk Factors and Non-Invasive Imaging (ORFAN) study

Abstract Background Patients with chest pain symptoms often undergo coronary CT angiography (CCTA) for the diagnosis of obstructive coronary artery disease (CAD) to guide revascularisation. However, acute cardiac events could occur in the absence of obstructive CAD. Further risk stratification of patients could potentially be useful by quantifying coronary inflammation using the Fat Attenuation Index (FAI) Score, and an artificial intelligence (AI)-assisted prognostic model that captures the inflammatory risk by integrating CCTA-derived metrics (including FAI Score and plaque burden) and the patient’s clinical risk factors. Purpose To evaluate the risk of cardiac events among patients without obstructive CAD, and the prognostic value of integrating coronary inflammation in risk stratifying this patient group. Methods From the Oxford Risk Factors and Non-invasive imaging (ORFAN) study, patients undergoing CCTA in 7 UK Hospitals (n=40,091) were followed up for the incidence of cardiac mortality over a median(IQR) of 2.7(1.4-5.3) years. In a nested cohort of 3,393 patients with long-term follow up, we evaluated the long-term prognostic value of FAI Score and the performance of AI-Risk algorithm over a median(IQR) follow up of 7.7(6.4-9.1) years. Reclassification of risk category from AI-Risk was compared with contemporary clinical risk score (QRISK3). Results Patients without obstructive CAD (n=32,533, 81.1%) accounted for 63.7% of all cardiac deaths (n=1,754) during follow up, highlighting the substantial proportion of cardiac events in absolute terms given a large proportion of patients without obstructive CAD. In the nested cohort with long-term follow up, FAI Score in any of the 3 epicardial coronary arteries was significantly associated with cardiac mortality with or without obstructive CAD. In the whole cohort, the highest quartile of FAI Score in LAD showed 20-times higher risk of cardiac death compared with the lowest quartile (Panel A). Integrating coronary inflammation with clinical risk factors and plaque burden, patients with very-high AI-Risk classification showed 6-fold higher risk of cardiac mortality compared to low/medium risk, with good alignment of the incident predicted events from AI-Risk and observed events. (Panel B) The predicted cardiac mortality from AI-Risk also demonstrated good calibration with the observed events in the whole population. (Panel C) Finally, AI-Risk significantly reclassify the risk above clinical risk factor-based prediction (QRISK3) (Panel D). Conclusion Patients with high coronary inflammation measured by FAI Score showed significantly elevated risk for cardiac mortality. The AI-enabled absolute risk prediction algorithm leads to significant reclassification of patients undergoing routine CCTA, and can be used as a precision medicine tool.

Prevecardio: a single-visit population screening model for detection and prevention of 10-year cardiovascular risk

Abstract Introduction Cardiovascular diseases (CVDs) are a major contributor to global mortality. The early detection of cardiovascular risk factors, coupled with timely intervention strategies, is essential for CVD prevention. PreveCardio was developed as an innovative, single-visit population screening model designed to facilitate the early identification and management of cardiovascular risk factors within a population of 1 million in a European capital city region totaling 6 million inhabitants. Purpose This study aimed to ascertain the prevalence of cardiovascular risk factors within the general population and to evaluate the effectiveness of the PreveCardio model in estimating 10-year cardiovascular risks and promoting preventive interventions. Methods In this cross-sectional prevalence study, individuals aged 50 to 75 from a health network were randomly selected and invited to participate in a comprehensive health check. The 30-minute screening consisted of lifestyle evaluations, anthropometric measurements, and blood tests for lipid profiles and glycated haemoglobin. The methodology was designed to provide participants with immediate feedback on their results, cardiovascular health and practical guidance for risk management. Results From the 8,492 individuals solicited, 3,546 took part in the screening: 56,83% were female, mean age 61,17 ± 7,33 y.o. Picture 1 shows the population's percentages with various modifiable cardiovascular risk factors, including 20% with active smoking, 30% with sedentarism, 16% with risky alcohol consumption, 49% with non-adherence to the Mediterranean diet, 40% with overweight, and 30% with obesity. The screening revealed new diagnoses of LDL-hypercholesterolemia in 47% participants, high levels of blood pressure in hypertension range in 49%, prediabetes in 30% and diabetes in 2% (Picture 2). After use SCORE2 and SCORE-OP risk prediction algorithms, PreveCardio is estimated to potentially prevent between 39 to 202 fatal and non-fatal cardiovascular events over 10 years. Predictive modelling based on PreveCardio data projected that expanding the screening across the capital region could prevent at least 26,260 cardiovascular events over 10 years. Conclusion(s) PreveCardio has established itself as a viable and effective model for the early detection of cardiovascular risk factors in a general population. It holds significant promise for public health impact by potentially reducing the incidence of CVD through comprehensive screening and proactive intervention.

Application of machine learning for predictions of consecutive dependent data of type {[(a, b)->c]->d}

Objective: Machine learning techniques have emerged in response to the desire for automatic pattern detection withindatasets in fields such as statistics, mathematics, and data analytics. They allow for the extraction of relevant informationfrom datasets of significantly large volumes, providing the possibility of making predictions. This paper presents an application focused on decision trees, linear regression, and random forest regression algorithms to predict final data fromconsecutive dependent data of type {[(a, b) → c] → D}. Methodology: The study adopts a quantitative research design, which takes as input datasets based on interval data. It utilizes a correlational research model by implementing Python and its Scikit-Learn library, which includes various algorithms for prediction. Specifically, we compare the application of decision trees, linear regression, and random forest regression on the same set of datasets, but with a characteristic of dependency between them. Results: Upon application of the proposed model, it yields an estimated prediction score, which indicates the accuracy of the model concerning the data provided. Conclusions: The application of a complex algorithm does not inherently guarantee a higher rate of accuracy. Conversely, configuring the model correctly, training multiple trees, or adjusting parameter values can significantly enhance the obtained results

The power of artificial intelligence for managing pandemics: A primer for public health professionals.

Artificial intelligence (AI) applications are complex and rapidly evolving, and thus often poorly understood, but have potentially profound implications for public health. We offer a primer for public health professionals that explains some of the key concepts involved and examines how these applications might be used in the response to a future pandemic. They include early outbreak detection, predictive modelling, healthcare management, risk communication, and health surveillance. Artificial intelligence applications, especially predictive algorithms, have the ability to anticipate outbreaks by integrating diverse datasets such as social media, meteorological data, and mobile phone movement data. Artificial intelligence-powered tools can also optimise healthcare delivery by managing the allocation of resources and reducing healthcare workers' exposure to risks. In resource distribution, they can anticipate demand and optimise logistics, while AI-driven robots can minimise physical contact in healthcare settings. Artificial intelligence also shows promise in supporting public health decision-making by simulating the social and economic impacts of different policy interventions. These simulations help policymakers evaluate complex scenarios such as lockdowns and resource allocation. Additionally, it can enhance public health messaging, with AI-generated health communications shown to be more effective than human-generated messages in some cases. However, there are risks, such as privacy concerns, biases in models, and the potential for 'false confirmations', where AI reinforces incorrect decisions. Despite these challenges, we argue that AI will become increasingly important in public health crises, but only if integrated thoughtfully into existing systems and processes.

Machine learning based prediction models for cardiovascular disease risk using electronic health records data: systematic review and meta-analysis

Abstract Background Cardiovascular disease (CVD) remains a major cause of mortality in the UK, prompting the need for improved risk predictive models for primary prevention. Machine learning (ML) models utilizing electronic health records (EHR) offer potential enhancements over traditional risk scores like QRISK3 and ASCVD. Objective To systematically evaluate and compare the efficacy of ML models against conventional CVD risk prediction algorithms using EHR data for medium to long-term (5-10 years) CVD risk prediction. Methods A systematic review and random-effect meta-analysis were conducted according to PRISMA guidelines, assessing studies from 2010 to 2024. We retrieved 32 ML models and 26 conventional statistical models from 20 selected studies, focusing on performance metrics such as Area Under the Curve (AUC) and heterogeneity across models. Results ML models, particularly random forest and deep learning, demonstrated superior performance, with the highest recorded pooled AUCs of 0.865 (95% CI: 0.812-0.917) and 0.847 (95% CI: 0.766-0.927), respectively. These significantly outperformed the conventional risk score of 0.765 (95% CI: 0.734-0.796). However, significant heterogeneity (I² > 99%) and potential publication bias were noted across the studies. Conclusion While ML models show enhanced calibration for CVD risk, substantial variability and methodological concerns limit their current clinical applicability. Future research should address these issues by enhancing methodological transparency and standardization to improve the reliability and utility of these models in clinical settings. Implications This study highlights the advanced capabilities of ML models in CVD risk prediction and emphasizes the need for rigorous validation to facilitate their integration into clinical practice. This review was registered with PROSPERO (CRD42021250058).

Comparison of transformer, LSTM and coupled algorithms for soil moisture prediction in shallow-groundwater-level areas with interpretability analysis

Accurate quantification of soil moisture is essential for understanding water and energy exchanges between the atmosphere and the Earth’s surface, as well as for agricultural applications. Predicting soil moisture content is vital for efficient water management, irrigation scheduling, and drought monitoring. Traditional forecasting methods, such as numerical regression models, often struggle due to various influencing factors and poor observation data quality. In contrast, deep learning algorithms, particularly recurrent and convolutional neural networks, show promise in predicting nonlinear data like soil moisture. This study focuses on shallow groundwater regions, using groundwater levels and meteorological data as features while coupling the Transformer model with other neural network structures. We investigate the potential of attention-based neural networks for soil moisture time series prediction. Our findings demonstrate that the Transformer model achieves an average R2 of 0.523 across different time lags, outperforming the LSTM model with an R2 of 0.485. The introduction of LSTM enhances the Transformer’s stability in handling temporal changes. Additionally, we verified the importance of groundwater for soil moisture prediction. This study introduces new methods for soil moisture prediction and offers new insights and recommendations for the development of artificial intelligence technology for soil moisture prediction.

Integrative analysis of bulk and single-cell transcriptomic data reveals novel insights into lipid metabolism and prognostic factors in hepatocellular carcinoma

Hepatocellular carcinoma (HCC) is associated with high mortality rate. This study investigated the status of lipid metabolism-related genes in HCC. Bulk transcriptomic and single-cell sequencing data for HCC were retrieved from public databases. The single-cell sequencing data was subjected to dimensionality reduction, which facilitated the annotation of distinct cell subpopulations and marker gene expression analysis within each subpopulation. Genes associated with lipid metabolism in liver cells were identified, and a machine-learning model was developed using the bulk transcriptomic data randomly partitioned into training and validation sets. The efficacy of the model was validated using these two sets. A multifactorial Cox analysis on the model genes combined with clinical features, led to the identification of age, HMGCS2, HNRNPU, and RAN as independent prognostic factors, which were included in the nomogram model construction and validation. A weighted gene co-expression analysis of all genes of the bulk transcriptome samples revealed the correlation between gene modules and risk score. Genes with cor > 0.4 in the highest-expressing module were selected for Gene Ontology and Kyoto Encyclopedia of Genes and Genomes functional enrichment analysis. Immune-related analysis was conducted based on seven algorithms for immune cell infiltration prediction. For the genes in the nomogram model, the expression in clinical pathological factors was also analyzed. The drug sensitivity analysis offered a reference for the selection of targeting drugs. This investigation provides novel insights and a theoretical basis for the prognosis, treatment, and pharmaceutical advancements for patients diagnosed with HCC.

Time-series water prediction based on feature clustering fusion and XGboost

Abstract. The time series prediction problem has a very wide range of applications in many fields. Most scholars use the LSTM class of algorithms for prediction. However, this method consumes a significant amount of computational power and presents several challenges. To address this issue, this paper proposes a time series forecasting method based on feature fusion and XGBoost. Specifically, we first utilize holiday information and the K-Means algorithm for feature extraction to expand the feature dimensions of the dataset, and then employ XGBoost as a model for training and prediction. Experiments demonstrate that the method proposed in this paper significantly reduces error compared to other traditional machine learning and deep learning methods, while the training time is much shorter than these methods. For example, compared with LSTM, the MSLE of this model decreases by 1.42%, while the training time is only 0.15% of that of LSTM. This greatly saves on training costs and computational power consumption. This confirms the effectiveness of using machine learning and clustering algorithms in time series prediction and provides new methods and practical application directions for future time series prediction models.

Identification of Key miRNAs in Endometriosis.

Endometriosis, a prevalent gynecological disorder characterized by the presence of endometrial-like tissue outside the uterus, poses significant challenges in diagnosis and management due to its unclear pathogenesis and lack of specific biomarkers. This study investigates the potential use of microRNAs (miRNAs) as key markers in endometriosis by studying two cohorts of patients (14 patients diagnosed with endometriosis and 15 patients with gynecological benign lesions, different from endometriosis). MicroRNA sequencing analysis was tested within data management by a custom pipeline designed by Eurofins Genoma Group. We identified a specific miRNA expression profile associated with endometriosis to feature specific disease molecular clusters to further elucidate the underlying mechanisms driving endometriosis pathogenesis. Data from the present study suggest a specific miRNA scar for endometriosis compared to other gynecological diseases to develop screening tools in early diagnosis and to ameliorate the management of the disease itself. This study lays the foundation for the identification of key miRNAs involved in the disease pathogenesis to unveil the molecular signatures in the complex scenario of endometriosis. Further validation and exploration of these findings are needed to develop tools to improve molecular diagnosis and to create a machine-learning prediction algorithm in the future.

Algorithms for Effector Prediction in Plant Pathogens and Pests: Achievements and Current Challenges

Algorithms for Effector Prediction in Plant Pathogens and Pests: Achievements and Current Challenges