Development Of Predictive Models Research Articles

Upfront surgery for intrahepatic cholangiocarcinoma: Prediction of futility using artificial intelligence

ObjectiveWe sought to identify patients at risk of “futile” surgery for intrahepatic cholangiocarcinoma using an artificial intelligence (AI)–based model based on preoperative variables. MethodsIntrahepatic cholangiocarcinoma patients who underwent resection between 1990 and 2020 were identified from a multi-institutional database. Futility was defined either as mortality or recurrence within 12 months of surgery. Various machine learning and deep learning techniques were used to develop prediction models for futile surgery. ResultsOverall, 827 intrahepatic cholangiocarcinoma patients were included. Among 378 patients (45.7%) who had futile surgery, 297 patients (78.6%) developed intrahepatic cholangiocarcinoma recurrence and 81 patients (21.4%) died within 12 months of surgical resection. An ensemble model consisting of multilayer perceptron and gradient boosting classifiers that used 10 preoperative factors demonstrated the highest accuracy, with areas under receiver operating characteristic curves of 0.830 (95% confidence interval 0.798–0.861) and 0.781 (95% confidence interval 0.707–0.853) in the training and testing cohorts, respectively. The model displayed sensitivity and specificity of 64.5% and 80.0%, respectively, with positive and negative predictive values of 73.1% and 72.7%, respectively. Radiologic tumor burden score, serum carbohydrate antigen 19-9, and direct bilirubin levels were the factors most strongly predictive of futile surgery. The artificial intelligence–based model was made available online for ease of use and clinical applicability (https://altaf-pawlik-icc-futilityofsurgery-calculator.streamlit.app/). ConclusionThe artificial intelligence ensemble model demonstrated high accuracy to identify patients preoperatively at high risk of undergoing futile surgery for intrahepatic cholangiocarcinoma. Artificial intelligence–based prediction models can provide clinicians with reliable preoperative guidance and aid in avoiding futile surgical procedures that are unlikely to provide patients long-term benefits.

Assessing the Capabilities of UV-NIR Spectroscopy for Predicting Macronutrients in Hydroponic Solutions with Single-Task and Multi-Task Learning

Macronutrients, including nitrogen (N), phosphorus (P), potassium (K), calcium (Ca), magnesium (Mg), and sulfur (S), are the most basic nutrient elements in the solution for the hydroponic system. However, the current management of hydroponic nutrient solutions usually depends on EC and pH sensors due to the lack of accurate specific macronutrient sensing equipment, which easily leads to nutritional imbalance for the cultivated plant. In this study, the UV-NIR absorption spectroscopy (200–1100 nm) was used to predict six macronutrients in hydroponic solutions; two kinds of single-task learning algorithms, including partial least squares (PLS) and least absolute shrinkage and selection operator (LASSO), and two kinds of multi-task learning algorithms, including dirty multi-task learning (DMTL) and robust multi-task learning (RMTL), were investigated to develop prediction models and assess capabilities of UV-NIR. The results showed that N and Ca could be quantitatively predicted by UV-NIR with the ratio of performance to deviation (RPD) more than 2, K could be qualitatively predicted (1.4 < RPD < 2), and P, Mg, and S could not be successfully predicted (RPD < 1.4); the RMTL algorithm outperformed others for predicting K and Ca benefit from the underlying task relationships with N; and predicting P, Mg, and S were identified as irrelevant (outlier) tasks. Our study provides a potential approach for predicting several macronutrients in hydroponic solutions with UV-NIR, especially using RMTL to improve model prediction ability.

Determinants and prediction of hypertension among Chinese middle-aged and elderly adults with diabetes: A machine learning approach

ObjectiveMultimorbidity, particularly diabetes combined with hypertension (DCH), is a significant public health concern. Currently, there is a gap in research utilizing machine learning (ML) algorithms to predict hypertension risk in Chinese middle-aged and elderly diabetic patients, and gender differences in DCH comorbidity patterns remain unclear. We aimed to use ML algorithms to predict DCH and identify its determinants among middle-aged and elderly diabetic patients in China. Study designCross-sectional study. MethodsData were collected on 2775 adults with diabetes aged ≥45 years from the 2015 China Health and Retirement Longitudinal Study. We employed nine ML algorithms to develop prediction models for DCH. The performance of these models was evaluated using the area under the curve (AUC). Additionally, we conducted variable importance analysis to identify key determinants. ResultsOur results showed that the best prediction models for the overall population, men, and women were extreme gradient boosting (AUC = 0.728), light gradient boosting machine (AUC = 0.734), and random forest (AUC = 0.737), respectively. Age, waist circumference, body mass index, creatinine level, triglycerides, taking Western medicine, high-density lipoprotein cholesterol, blood urea nitrogen, total cholesterol, low-density lipoprotein cholesterol, and sleep disorders were identified as common important predictors by all three populations. ConclusionsML algorithms showed accurate predictive capabilities for DCH. Overall, non-linear ML models outperformed traditional logistic regression for predicting DCH. DCH predictions exhibited variations in predictors and model accuracy by gender. These findings could help identify DCH early and inform the development of personalized intervention strategies.

An Automated Malnutrition Screening Tool Using Routinely Collected Data for Older Adults in Long-Term Care: Development and Internal Validation of AutoMal

ObjectiveTo develop and internally validate a malnutrition screening tool based on routinely collected data in the long-term care setting. DesignDiagnostic prediction model development and internal validation study. Setting and ParticipantsResidents (n = 539) from 10 long-term care facilities in Australia. MethodsCandidate variables identified through expert consultation were collected from routinely collected data in a convenience sample of long-term care facilities. Logistic regression using the Subjective Global Assessment as the reference standard was conducted on 500 samples derived using bootstrapping from the original sample. Candidate variables were selected if included in more than 95% of samples using backwards stepwise elimination. The final model was developed using logistic regression of selected variables. Internal validation was conducted using bootstrapping to calculate the optimism-adjusted performance. Overall discrimination was evaluated via receiver operator characteristic curve and calculation of the area under the curve. Youden's Index was used to identify the optimal threshold value for classifying malnutrition. Sensitivity and specificity were calculated. ResultsBody mass index and weight change % over 6 months were included in the automated malnutrition screening model (AutoMal), identified in 100% of bootstrapped samples. AutoMal demonstrated excellent discrimination of malnutrition, with area under the curve of 0.8378 (95% CI, 0.80–0.87). Youden's Index value was 0.37, resulting in sensitivity of 78% (95% CI, 71%–83%) and specificity of 77% (72%–81%). Optimism-corrected area under the curve was 0.8354. Conclusions and ImplicationsThe AutoMal demonstrates excellent ability to differentiate malnutrition status. It makes automated identification of malnutrition possible by using 2 variables commonly found in electronic health records.

Prediction models for postpartum stress urinary incontinence: A systematic review

BackgroundPostpartum stress urinary incontinence significantly impacts the quality of life and the physical and mental health of women. A reliable predictive model for postpartum stress urinary incontinence can serve as a preventive tool. Currently, there have been numerous studies developing predictive models to assess the risk of postpartum stress urinary incontinence, but the quality and clinical applicability of these models remain unknown. ObjectiveTo systematically review and evaluate existing models for predicting stressful postpartum risks. MethodsPubMed, EBSCO, The Cochrane Library, Embase, Web of Science, China National Knowledge Infrastructure, WanFang Data, SinoMed, and VIP Data databases were systematically searched from the time of database construction to October 2023. Two researchers used Critical appraisal and data extraction for systematic reviews of prediction modeling studies: the CHARMS checklist for data extraction. Three researchers used The Prediction Model Risk of Bias Assessment Tool (PROBAST) checklist for bias and applicability assessment. ResultsEight papers including ten postpartum stress urinary incontinence prediction models were finalized. The most common predictors in the prediction models were urinary incontinence (UI) during pregnancy, followed by mode of delivery, Maternal age, parity, and UI before pregnancy. Nine of the prediction models reported discrimination with an area under the ROC curve (AUC) or C-index between 0.680 and 0.850. All included studies were at high risk of bias, mainly due to mishandling of continuous predictors, unreported or mishandled missing data, and inadequate assessment of predictive model performance. ConclusionsPostpartum stress urinary incontinence risk prediction models are in the initial development stage, and existing prediction models have a high risk of bias and poor modeling methodological quality, which may hinder their clinical application. In the future, healthcare practitioners should follow the norms of predictive model development and reporting to develop risk prediction models with superior predictive performance, low risk of bias, and easy clinical application.

Penerapan Algoritma Naïve Bayes Untuk Klasifikasi Status Gizi Stunting Pada Balita

Child stunting is a major public health concern in Indonesia. This study uses the Naïve Bayes classification algorithm to assess the nutritional condition of stunted children based on demographic and anthropometric characteristics. The information used comes from the Toddler Weighing Month (Bulan Penimbangan Balita - BPB) in Majalengka Regency. Data type conversion, separating data into training and testing sets, and data normalization are all examples of preprocessing steps. The model's evaluation results reveal an accuracy of 94.65%, with precision and recall for each category of stunted nutritional status. This study makes a substantial contribution to early diagnosis and mitigation of stunting in Indonesia, as well as providing the framework for future development of more powerful predictive models.

Inteligencia Artificial, Imagen Médica y Medicina de Precisión: Avances y Perspectivas

The role of artificial intelligence (AI) applied to medical imaging in developing and strengthening personalized medicine is described as a continuous process of improvement, a set of opportunities, and a professional challenge of enormous significance. This paper outlines the main processes in which AI is involved regarding imaging, including data preparation, image harmonization, automatic segmentation of organs and lesions, labeling, extraction of radiomic features, and the development of predictive clinical models. It will also address aspects related to the integration of these solutions into clinical practice to enhance accuracy and efficiency in the care process, diagnosis, and treatment, making it more personalized, efficient, and precise. Projects like PRIMAGE and CHAIMELEON highlight the transformative potential of AI and the fundamental role of interdisciplinary collaboration in realizing this potential, based on ongoing multiprofesional collaboration to address the ethical, regulatory, technical, and clinical challenges that accompany these advancements.

Development and Verification of a Fire Occurrence Prediction Model with Social-Architectural Factors

Development and Verification of a Fire Occurrence Prediction Model with Social-Architectural Factors

A Comprehensive Review on : Aquaponic Farming Water Quality Prediction

Aquaponic farming, which combines aquaculture and hydroponics, depends strongly on maintaining optimal water quality to guarantee the health and productivity of both fish and plants. This review paper explores the latest developments in IoT-based automated water monitoring systems, focusing on their role in predicting and managing water quality in aquaponic systems. Regardless of significant progress there are several research gaps. Recent studies highlight challenges such as inconsistent sensor selection, calibration issues, insufficient publicly available data, and inadequate data cleaning and preprocessing. Also, the issues of imbalanced datasets, limited long-term data, and underdeveloped IoT and AI integration prevent the development of accurate predictive models. The scalability and maintenance of systems, understanding microbial dynamics, and nutrient management are also critical areas needing further exploration. This review also identifies the need for deeper case studies and advanced feature extraction methods to enhance prediction accuracy. By addressing these gaps, including system scalability and nutrient management, future research can improve data availability and quality, enabling more robust predictions and contributing to more efficient and sustainable aquaponic systems.

Neutrophil estimation and prognosis analysis based on existing lung squamous cell carcinoma datasets: the development and validation of a prognosis prediction model.

Notwithstanding the rapid developments in precision medicine in recent years, lung cancer still has a low survival rate, especially lung squamous cell cancer (LUSC). The tumor microenvironment (TME) plays an important role in the progression of lung cancer, in which high neutrophil levels are correlated with poor prognosis, potentially due to their interactions with tumor cells via pro-inflammatory cytokines and chemokines. However, the precise mechanisms of how neutrophils influence lung cancer remain unclear. This study aims to explore these mechanisms and develop a prognosis predictive model in LUSC, addressing the knowledge gap in neutrophil-related cancer pathogenesis. LUSC datasets from the Xena Hub and Gene Expression Omnibus (GEO) databases were used, comprising 473 tumor samples and 195 tumor samples, respectively. Neutrophil contents in these samples were estimated using CIBERSORT, xCell, and microenvironment cell populations (MCP) counter tools. Differentially expressed genes (DEGs) were identified using DEseq2, and a weighted gene co-expression network analysis (WGCNA) was performed to identify neutrophil-related genes. A least absolute shrinkage and selection operator (LASSO) Cox regression model was constructed for prognosis prediction, and the model's accuracy was validated using Kaplan-Meier survival curves and time-dependent receiver operating characteristic (ROC) curves. Additionally, genomic changes, immune correlations, drug sensitivity, and immunotherapy response were analyzed to further validate the model's predictive power. Neutrophil content was significantly higher in adjacent normal tissue compared to LUSC tissue (P<0.001). High neutrophil content was associated with worse overall survival (OS) (P=0.02), disease-free survival (DFS) (P=0.02), and progression-free survival (PFS) (P=0.03) using different software estimates. Nine gene modules were identified, with blue and yellow modules showing strong correlations with neutrophil prognosis (P<0.001). Eight genes were selected for the prognostic model, which accurately predicted 1-, 3-, and 5-year survival in both the training set [area under the curve (AUC) value =0.60, 0.63, 0.66, respectively] and validation set (AUC value =0.58, 0.58, 0.59, respectively), with significant prognosis differences between high- and low-risk groups (P<0.001). The model's independent prognostic factors included risk group, pathologic M stage, and tumor stage (P<0.05). A further molecular mechanism analysis revealed differences between risk groups were revealed in immune checkpoint and human leukocyte antigen (HLA) gene expression, hallmark pathways, drug sensitivity, and immunotherapy responses. This study established a risk-score model that effectively predicts the prognosis of LUSC patients and sheds light on the molecular mechanisms involved. The findings enhance the understanding of neutrophil-tumor interactions, offering potential targets for personalized treatments. However, further experimental validation and clinical studies are required to confirm these findings and address study limitations, including reliance on public databases and focus on a specific lung cancer subtype.

Proactive cyber threat mitigation: Integrating data-driven insights with user-centric security protocols

In the rapidly evolving cybersecurity landscape, proactive threat mitigation has become essential for protecting sensitive data and systems from increasingly sophisticated attacks. The paper explores a holistic approach to enhancing cybersecurity by merging data-driven insights with user-centric security practices. This study highlights the limitations of traditional reactive security measures and advocates for a shift towards a proactive strategy that anticipates and addresses potential threats before they materialize. The research emphasizes the importance of leveraging data analytics to gain actionable insights into potential vulnerabilities and emerging threats. By analyzing large datasets from network traffic, user behavior, and historical incidents, organizations can identify patterns and anomalies that signal potential risks. These data-driven insights enable the development of more effective threat detection systems and predictive models that enhance an organization’s ability to respond swiftly to evolving cyber threats. In addition to data analytics, the study underscores the significance of integrating user-centric security protocols. Traditional security measures often overlook the human element, which can be a critical factor in both the efficacy of security measures and the overall risk profile. User-centric protocols focus on enhancing user awareness, implementing tailored security training, and promoting best practices in cybersecurity. This approach ensures that security measures are not only technologically sound but also aligned with user behavior and needs, reducing the risk of human error and improving overall security posture. The research also discusses practical implementation strategies for combining data-driven insights with user-centric protocols, including the use of advanced analytics tools, continuous user education, and adaptive security frameworks. Case studies demonstrate the benefits of this integrated approach, such as improved threat detection rates, reduced incident response times, and enhanced overall security resilience. In conclusion, this study presents a comprehensive framework for proactive cyber threat mitigation, emphasizing the synergy between data-driven analytics and user-centric security practices as a robust strategy for safeguarding against modern cyber threats. Keywords: Cyber Threat, Proactive, Data-Driven, User-Centric, Security Protocols

A model for energy predictions and diagnostics of large-scale photovoltaic systems based on electric data and thermal imaging of the PV fields

The aim of this investigation is the development of robust models for the performance prediction and automatic monitoring of large photovoltaic systems, based on historical and real-time electric and thermal data. This issue is increasingly important due to the worldwide diffusion of large photovoltaic systems and their need to identify and predict failures and malfunctions, in order to promptly assess the convenience of maintenance actions. The present model describes the response to irradiance and temperature conditions of both modules and inverters and also it is able to predict shading conditions able to affect the energy yield. The model has been validated against real electric measurements in 6 large PV plants located in southern Italy and it demonstrated to be able to predict the real time power production within a 4.1 % error. Even more importantly, the model and its comparison with subhourly measurements over several years has demonstrated its effectiveness in detecting downtime conditions caused by inverter or string problems. Simulations and measurements revealed that missed energy production due to electrical grid coupling downtime can exceed 50 % on certain days and that the shading conditions (up to 5 % of the daily energy production) can be easily detected and separated from component problems, thus avoiding false alarms. Finally, the analysis of aerial infrared images allowed to further test the model in failure detection capability, assess the relationship between thermal anomalies and underperformance conditions and in predicting the yearly deterioration rate at the PV plants.

819: Development of an explainable AI Model for OS prediction in LA-NSCLC patients receiving RT

819: Development of an explainable AI Model for OS prediction in LA-NSCLC patients receiving RT

1505: Development of an AI predictive model for patient selection for adaptive head and neck radiotherapy.

1505: Development of an AI predictive model for patient selection for adaptive head and neck radiotherapy.

A case study of multi-objective design optimization of a healthy building in Shanghai, China

Energy efficient healthy buildings design is important in achieving carbon neutrality and occupants’ health, yet not sufficiently explored. This paper aims to optimize a healthy building in Shanghai, China, based on energy consumption, indoor air quality and visual comfort. A four-step optimization method was proposed. Firstly, Latin Hypercube-Soubert sampling method was used to generate 375 design samples and performance data through computer simulation and calculation. Secondly, five different machine learning approaches were applied for prediction model development. Thirdly, the best prediction models were coupled with eleven optimization algorithms to find Pareto solutions. Finally, optimization on different combinations of design parameters were conducted. It was found that the back propagation neural network and Pareto Envelope-based Selection Algorithm II were the best prediction models and optimization algorithm. The average reduction of building energy consumption, visual discomfort, and improper indoor air quality hours were found to be 25.73 %, 46.24 %, and 38.34 %, respectively. The recommended windows to wall ratios of the east, south, west and north walls, absorptance of solar radiation and filter types are in the range of 20%–35 %, 10%–45 %, 10%–30 %, 10%–40 %, 0.7–0.9, and S7-S9, respectively. This study contributes to enrich the case study of healthy buildings, provide a quick design optimization approach and analysis on the importance of each design parameter. Its originality lies in the optimization methodology, comparative analysis of prediction models, optimization algorithms, and combination effects of design parameters. The outcomes can guide the building designers in performing energy efficient design while maximizing indoor air quality and visual comfort.

Prediction of Glass Transition Temperature of Polymers Using Simple Machine Learning.

Polymer materials have garnered significant attention due to their exceptional mechanical properties and diverse industrial applications. Understanding the glass transition temperature (Tg) of polymers is critical to prevent operational failures at specific temperatures. Traditional methods for measuring Tg, such as differential scanning calorimetry (DSC) and dynamic mechanical analysis, while accurate, are often time-consuming, costly, and susceptible to inaccuracies due to random and uncertain factors. To address these limitations, the aim of the present study is to investigate the potential of Simplified Molecular Input Line Entry System (SMILES) as descriptors in simple machine learning models to predict Tg efficiently and reliably. Five models were utilized: k-nearest neighbors (KNNs), support vector regression (SVR), extreme gradient boosting (XGBoost), artificial neural network (ANN), and recurrent neural network (RNN). SMILES descriptors were converted into numerical data using either One Hot Encoding (OHE) or Natural Language Processing (NLP). The study found that SMILES inputs with fewer than 200 characters were inadequate for accurately describing compound structures, while inputs exceeding 200 characters diminished model performance due to the curse of dimensionality. The ANN model achieved the highest R2 value of 0.79; however, the XGB model, with an R2 value of 0.774, exhibited the highest stability and shorter training times compared to other models, making it the preferred choice for Tg prediction. The efficiency of the OHE method over NLP was demonstrated by faster training times across the KNN, SVR, XGB, and ANN models. Validation of new polymer data showed the XGB model's robustness, with an average prediction deviation of 9.76 from actual Tg values. These findings underscore the importance of optimizing SMILES conversion methods and model parameters to enhance prediction reliability. Future research should focus on improving model accuracy and generalizability by incorporating additional features and advanced techniques. This study contributes to the development of efficient and reliable predictive models for polymer properties, facilitating the design and application of new polymer materials.

Machine learning predictive insight of water pollution and groundwater quality in the Eastern Province of Saudi Arabia

This study presents an innovative approach for predicting water and groundwater quality indices (WQI and GWQI) in the Eastern Province of Saudi Arabia, addressing critical challenges of scarcity and pollution in arid regions. Recent literature highlights the increasing attention towards WQI based on water pollution index (WPI) and GWQI as essential tools for simplifying complex hydrogeological data, thereby facilitating effective groundwater management and protection. Unlike previous works, the present research introduces a novel hybrid method that integrates non-parametric kernel Gaussian learning (GPR), adaptive neuro-fuzzy inference system (ANFIS), and decision tree (DT) algorithms. This approach marks the first application of a non-parametric kernel for groundwater quality pollution index prediction in Saudi Arabia, offering a significant advancement in the field. Through laboratory analysis and the combination of various machine learning (ML) techniques, this study enhances prediction capabilities, particularly for unmonitored sites in arid and semi-arid regions. The study’s objectives include feature engineering based on dependency sensitivity analysis to identify the most influential variables affecting WQI and GWQI, and the development of predictive models using ANFIS, GPR, and DT for both indices. Furthermore, it aims to assess the impact of different data portions on WQI and GWQI predictions, exploring data divisions such as (70% / 30%), (60% / 40%), and (80% / 20%) for training and testing phase, respectively. By filling a critical gap in water resource management, this research offers significant implications for the prediction of water quality in regions facing similar environmental challenges. Through its innovative methodology and comprehensive analysis, this study contributes to the broader effort of managing and protecting water resources in arid and semi-arid areas. The result proved that GPR-M1 exhibited exceptional testing phase accuracy with RMSE = 0.0169 for GWQI. Similarly, for WPI, the ANFIS-M1 achieved high testing predictive skills with RMSE = 0.0401. The results emphasize the critical role of data quality and quantity in training for enhancing model robustness and prediction precision in water quality assessment.

Kinetic study of the thermal inactivation of Weizmannia coagulans during food thermal processing

Weizmannia coagulans has attracted attention due to its remarkable health benefits for human, but the dynamic changes of its viable bacteria during thermal processing have been less reported. In this study, a predictive model for the survival of Weissmanella coagulans during thermal processing of food was developed and validated during the processing of coffee, tea, instant noodles, calcium milk biscuits, muffin cake and steamed buns. The kinetics of heat inactivation activities of Weizmannia coagulans VHProbi C08 and Weizmannia coagulans GBI-30, 6086 at 85, 95, 105, 110 and 115 °C were investigated, and their coefficients of determination were greater than 0.91 and 0.87, and the root-mean-square errors were less than 0.64 and 0.43, respectively. The z-values of VHProbi C08 and GBI-30, 6086 were obtained by Bigelow model fitting as 36.1 °C and 36.9 °C, respectively. The developed prediction model was applied to the thermal processing of six food products and the measured values were all within ±0.5 Log10 (CFU/mL) of the predicted values, indicating high prediction accuracy. The model predicts the survival of Weissmanella coagulans simply by obtaining the initial number of viable bacteria and the change in temperature. These suggested that the model can be used as an effective tool to evaluate the stability of Weizmannia coagulans in food thermal processing.

Innovative nomogram for predicting 1-year clinically driven revascularization outcomes in endovascular femoropopliteal disease.

Femoropopliteal artery disease (FPAD) is a common vascular disease that usually requires surgical treatment. The aim of this study was to apply predictive modeling to develop predictive models for predicting clinically driven target revascularization (CD-TLR) events 1 year after intervention in patients with FPAD. In this study, clinical data were collected from a total of 484 patients who underwent FPAD endovascular intervention from 2014 to 2019. According to the inclusion and exclusion criteria, 400 patients will be finally included and assigned to the training cohort and test cohort in the ratio of 7:3. By analyzing these data through statistical methods, we will explore the effects of different factors on target revascularization events 1 year after intervention in FPAD patients, and build the corresponding prediction model of the column line graph. The final nomogram model consisted of 5 independent predictors: history of cerebrovascular disease, lesion length >15 cm, no atherectomy device used, no medicated balloon used and procedure time. The C-index of the model was 0.766 and 0.726 for the training cohort and validation cohort, respectively. The calibration curves also showed that the model had satisfactory agreement in both cohorts. The newly developed prediction model can accurately predict clinically driven target revascularization events at 1 year in patients with FPAD, providing valuable information for the development of individualized treatment plans.

Development of a biomarker prediction model for post-trauma multiple organ failure/dysfunction syndrome based on the blood transcriptome

BackgroundMultiple organ failure/dysfunction syndrome (MOF/MODS) is a major cause of mortality and morbidity among severe trauma patients. Current clinical practices entail monitoring physiological measurements and applying clinical score systems to diagnose its onset. Instead, we aimed to develop an early prediction model for MOF outcome evaluated soon after traumatic injury by performing machine learning analysis of genome-wide transcriptome data from blood samples drawn within 24 h of traumatic injury. We then compared its performance to baseline injury severity scores and detection of infections.MethodsBuffy coat transcriptome and linked clinical datasets from blunt trauma patients from the Inflammation and the Host Response to Injury Study (“Glue Grant”) multi-center cohort were used. According to the inclusion/exclusion criteria, 141 adult (age ≥ 16 years old) blunt trauma patients (excluding penetrating) with early buffy coat (≤ 24 h since trauma injury) samples were analyzed, with 58 MOF-cases and 83 non-cases. We applied the Least Absolute Shrinkage and Selection Operator (LASSO) and eXtreme Gradient Boosting (XGBoost) algorithms to select features and develop models for MOF early outcome prediction.ResultsThe LASSO model included 18 transcripts (AUROC [95% CI]: 0.938 [0.890–0.987] (training) and 0.833 [0.699–0.967] (test)), and the XGBoost model included 41 transcripts (0.999 [0.997–1.000] (training) and 0.907 [0.816–0.998] (test)). There were 16 overlapping transcripts comparing the two panels (0.935 [0.884–0.985] (training) and 0.836 [0.703–0.968] (test)). The biomarker models notably outperformed models based on injury severity scores and sex, which we found to be significantly associated with MOF (APACHEII + sex—0.649 [0.537–0.762] (training) and 0.493 [0.301–0.685] (test); ISS + sex—0.630 [0.516–0.744] (training) and 0.482 [0.293–0.670] (test); NISS + sex—0.651 [0.540–0.763] (training) and 0.525 [0.335–0.714] (test)).ConclusionsThe accurate assessment of MOF from blood samples immediately after trauma is expected to aid in improving clinical decision-making and may contribute to reduced morbidity, mortality and healthcare costs. Moreover, understanding the molecular mechanisms involving the transcripts identified as important for MOF prediction may eventually aid in developing novel interventions.