Hyperparameter Search Research Articles

Recurrent Neural Networks for Artifact Correction in HRV Data During Physical Exercise


 
 
 In this paper, we propose the use of recurrent neural networks (RNNs) for artifact correction and analysis of heart rate variability (HRV) data. HRV can be a valuable metric for determining the function of the heart and the autonomic nervous system. When measured during exercise, motion artifacts present a significant challenge. Several methods for artifact correction have previously been proposed, none of them applying machine learning, and each presenting some limitations regarding an accurate representation of HRV metrics. RNNs offer the ability to capture patterns that might otherwise not be detected, yielding predictions where no prior physiological assumptions are needed.
 A hyperparameter search has been carried out to determine the best network configuration and the most important hyperparameters. The approach was tested on two extensive multi-subject data sets, one from a recreational bicycle race and the other from a laboratory experiment. The results demonstrate that RNNs outperform by order of magnitude existing methods with respect to the calculation of derived HRV metrics. However, they are not able to accurately fill in individual missing RR intervals in sequence. Future research should pursue improvements in the prediction of RR interval lengths and reduction in necessary training data.
 
 

SigOpt Mulch: An intelligent system for AutoML of gradient boosted trees

Gradient boosted trees (GBTs) are ubiquitous models used by researchers, machine learning (ML) practitioners, and data scientists because of their robust performance, interpretable behavior, and ease-of-use. One critical challenge in training GBTs is the tuning of their hyperparameters. In practice, selecting these hyperparameters is often done manually. Recently, the ML community has advocated for tuning hyperparameters through black-box optimization and developed state-of-the-art systems to do so. However, applying such systems to tune GBTs suffers from two drawbacks. First, these systems are not model-aware, rather they are designed to apply to a generic model; this leaves significant optimization performance on the table. Second, using these systems requires domain knowledge such as the choice of hyperparameter search space, which is an antithesis to the automatic experimentation that black-box optimization aims to provide. In this paper, we present SigOpt Mulch, a model-aware hyperparameter tuning system specifically designed for automated tuning of GBTs that provides two improvements over existing systems. First, Mulch leverages powerful techniques in metalearning and multifidelity optimization to perform model-aware hyperparameter optimization. Second, it automates the process of learning performant hyperparameters by making intelligent decisions about the optimization search space, thus reducing the need for user domain knowledge. These innovations allow Mulch to identify good GBT hyperparameters far more efficiently—and in a more seamless and user-friendly way—than existing black-box hyperparameter tuning systems.

The value of a machine learning algorithm to predict adverse short-term outcome during resuscitation of patients with in-hospital cardiac arrest: a retrospective study.

Background and importance Guidelines recommend that hospital emergency teams locally validate criteria for termination of cardiopulmonary resuscitation in patients with in-hospital cardiac arrest (IHCA). Objective To determine the value of a machine learning algorithm to predict failure to achieve return of spontaneous circulation (ROSC) and unfavourable functional outcome from IHCA using only data readily available at emergency team arrival. Design Retrospective cohort study. Setting and participants Adults who experienced an IHCA were attended to by the emergency team. Outcome measures and analysis Demographic and clinical data typically available at the arrival of the emergency team were extracted from the institutional IHCA database. In addition, outcome data including the Cerebral Performance Category (CPC) score count at hospital discharge were collected. A model selection procedure for random forests with a hyperparameter search was employed to develop two classification algorithms to predict failure to achieve ROSC and unfavourable (CPC 3-5) functional outcomes. Main results Six hundred thirty patients were included, of which 390 failed to achieve ROSC (61.9%). The final classification model to predict failure to achieve ROSC had an area under the receiver operating characteristic curve of 0.9 [95% confidence interval (CI), 0.89-0.9], a balanced accuracy of 0.77 (95% CI, 0.75-0.79), an F1-score of 0.78 (95% CI, 0.76-0.79), a positive predictive value of 0.88 (0.86-0.91), a negative predictive value of 0.61 (0.6-0.63), a sensitivity of 0.69 (0.66-0.72), and a specificity of 0.84 (0.8-0.88). Five hundred fifty-nine subjects experienced an unfavourable outcome (88.7%). The final classification model to predict unfavourable functional outcomes from IHCA at hospital discharge had an area under the receiver operating characteristic curve of 0.93 (95% CI, 0.92-0.93), a balanced accuracy of 0.59 (95% CI, 0.57-0.61), an F1-score of 0.94 (95% CI, 0.94-0.95), a positive predictive value of 0.91 (0.9-0.91), a negative predictive value of 0.57 (0.48-0.66), a sensitivity of 0.98 (0.97-0.99), and a specificity of 0.2 (0.16-0.24). Conclusion Using data readily available at emergency team arrival, machine learning algorithms had a high predictive power to forecast failure to achieve ROSC and unfavourable functional outcomes from IHCA while cardiopulmonary resuscitation was still ongoing; however, the positive predictive value of both models was not high enough to allow for early termination of resuscitation efforts.

ForeTiS: A comprehensive time series forecasting framework in Python

Summary: Time series forecasting is a research area with applications in various domains, nevertheless without yielding a predominant method so far. We present ForeTiS, a comprehensive and open source Python framework that allows rigorous training, comparison, and analysis of state-of-the-art time series forecasting approaches. Our framework includes fully automated yet configurable data preprocessing and feature engineering. In addition, we use advanced Bayesian optimization for automatic hyperparameter search. ForeTiS is easy to use, even for non-programmers, requiring only a single line of code to apply state-of-the-art time series forecasting. Various prediction models, ranging from classical forecasting approaches to machine learning techniques and deep learning architectures, are already integrated. More importantly, as a key benefit for researchers aiming to develop new forecasting models, ForeTiS is designed to allow for rapid integration and fair benchmarking in a reliable framework. Thus, we provide a powerful framework for both end users and forecasting experts.Availability:ForeTiS is available at https://github.com/grimmlab/ForeTiS. We provide a setup using Docker, as well as a Python package at https://pypi.org/project/ForeTiS/. Extensive online documentation with hands-on tutorials and videos can be found at https://foretis.readthedocs.io.

Genie in the Model

Advances in deep neural networks (DNNs) have fostered a wide spectrum of intelligent mobile applications ranging from voice assistants on smartphones to augmented reality with smart-glasses. To deliver high-quality services, these DNNs should operate on resource-constrained mobile platforms and yield consistent performance in open environments. However, DNNs are notoriously resource-intensive, and often suffer from performance degradation in real-world deployments. Existing research strives to optimize the resource-performance trade-off of DNNs by compressing the model without notably compromising its inference accuracy. Accordingly, the accuracy of these compressed DNNs is bounded by the original ones, leading to more severe accuracy drop in challenging yet common scenarios such as low-resolution, small-size, and motion-blur. In this paper, we propose to push forward the frontiers of the DNN performance-resource trade-off by introducing human intelligence as a new design dimension. To this end, we explore human-in-the-loop DNNs (H-DNNs) and their automatic performance-resource optimization. We present H-Gen, an automatic H-DNN compression framework that incorporates human participation as a new hyperparameter for accurate and efficient DNN generation. It involves novel hyperparameter formulation, metric calculation, and search strategy in the context of automatic H-DNN generation. We also propose human participation mechanisms for three common DNN architectures to showcase the feasibility of H-Gen. Extensive experiments on twelve categories of challenging samples with three common DNN structures demonstrate the superiority of H-Gen in terms of the overall trade-off between performance (accuracy, latency), and resource (storage, energy, human labour).

Chaotic Whale Optimization Algorithm in Hyperparameter Selection in Convolutional Neural Network Algorithm

In several previous studies, metaheuristic methods were used to search for CNN hyperparameters. However, this research only focuses on searching for CNN hyperparameters in the type of network architecture, network structure, and initializing network weights. Therefore, in this article, we only focus on searching for CNN hyperparameters with network architecture type, and network structure with additional regularization. In this article, the CNN hyperparameter search with regularization uses CWOA on the MNIST and FashionMNIST datasets. Each dataset consists of 60,000 training data and 10,000 testing data. Then during the research, the training data was only taken 50% of the total data, then the data was divided again by 10% for data validation and the rest for training data. The results of the research on the MNIST CWOA dataset have an error value of 0.023 and an accuracy of 99.63. Then the FashionMNIST CWOA dataset has an error value of 0.23 and an accuracy of 91.36.

Stress Evaluation in Rails Based on Vibration Data and Artificial Intelligence

An inspection technique for the estimation of longitudinal stress and rail neutral temperature (RNT) in continuous welded rails (CWR) is presented. The technique is based on the use of finite element analysis (FEA), vibration measurements, and machine learning (ML). The FEA is used to establish the dependency of the vibration characteristics (mode shapes and frequencies) of any given CWR to the conditions of the track and the longitudinal stress of the rail. The dependency is then used to train an ML algorithm that is tested using experimental data obtained with two accelerometers bonded to the rail of interest, during a field test conducted in Colorado, U.S. A commercial finite element (FE) software was used to model the rail track as a short segment repeated indefinitely and under varying boundary conditions and stress. Three ML models were developed using hyperparameter search optimization techniques and k-fold cross-validation to infer the stress or RNT from the frequencies of vibration extracted from the time waveforms recorded by the accelerometers. The results demonstrated that the success of the technique is highly dependent on the accuracy of the FE model. The results also demonstrated that ML was mostly able to learn from the experimental data and successfully predict the neutral temperature.

Data-driven predictive prognostic model for power batteries based on machine learning

Under the pressure of energy and environmental protection, new energy vehicles have become the future direction of automotive development. However, the safety performance of the power battery has always been the most critical indicator in the new energy vehicle industry. The battery will be aged in the continuous charging and discharging cycle, and the aging will cause safety hazards when it reaches a limit. A model that can predict the battery life can be obtained using Machine Learning. To obtain models that can predict power battery life relatively accurately, this paper revolves around the chaos sparrow search optimization algorithm, Random Forest, XGBoost, LightGBM, CatBoost, and NN, the importance assessment of the features, the hyperparameter search process, and the comparison of the differences and performance between the different algorithms are discussed. CatBoost has the highest prediction accuracy, with the amount of predicted data with a relative error of less than 10% being 88.44%. (a total of 10,275 data in the test set). And finally comes up with a general approach to predicting power battery life using Machine Learning.

The Need for Speed: A Fast Guessing Entropy Calculation for Deep Learning-Based SCA

The adoption of deep neural networks for profiling side-channel attacks opened new perspectives for leakage detection. Recent publications showed that cryptographic implementations featuring different countermeasures could be broken without feature selection or trace preprocessing. This success comes with a high price: an extensive hyperparameter search to find optimal deep learning models. As deep learning models usually suffer from overfitting due to their high fitting capacity, it is crucial to avoid over-training regimes, which require a correct number of epochs. For that, early stopping is employed as an efficient regularization method that requires a consistent validation metric. Although guessing entropy is a highly informative metric for profiling side-channel attacks, it is time-consuming, especially if computed for all epochs during training, and the number of validation traces is significantly large. This paper shows that guessing entropy can be efficiently computed during training by reducing the number of validation traces without affecting the efficiency of early stopping decisions. Our solution significantly speeds up the process, impacting the performance of the hyperparameter search and overall profiling attack. Our fast guessing entropy calculation is up to 16× faster, resulting in more hyperparameter tuning experiments and allowing security evaluators to find more efficient deep learning models.

Potential of machine learning methods in operational risk stratification in patients with coronary artery disease scheduled for coronary bypass surgery

Aim. To develop and evaluate the effectiveness of models for predicting mortality after coronary bypass surgery, obtained using machine learning analysis of preoperative data.Material and methods. As part of a cohort study, a retrospective prediction of in-hospital mortality after coronary artery bypass grafting (CABG) was performed in 2182 patients with stable coronary artery disease. Patients were divided into 2 following samples: learning (80%, n=1745) and training (20%, n=437). The initial ratio of surviving (n=2153) and deceased (n=29) patients in the total sample indicated a pronounced class imbalance, and therefore the resampling method was used in the training sample. Five machine learning (ML) algorithms were used to build predictive risk models: Logistic regression, Random Forrest, CatBoost, LightGBM, XGBoost. For each of these algorithms, cross-validation and hyperparameter search were performed on the training sample. As a result, five predictive models with the best parameters were obtained. The resulting predictive models were applied to the learning sample, after which their performance was compared in order to determine the most effective model.Results. Predictive models implemented on ensemble classifiers (CatBoost, LightGBM, XGBoost) showed better results compared to models based on logistic regression and random forest. The best quality metrics were obtained for CatBoost and LightGBM based models (Precision — 0,667, Recall — 0,333, F1-score — 0,444, ROC AUC — 0,666 for both models). There were following common high-ranking parameters for deciding on the outcome for both models: creatinine and blood glucose levels, left ventricular ejection fraction, age, critical stenosis (>70%) of carotid arteries and main lower limb arteries.Conclusion. Ensemble machine learning methods demonstrate higher predictive power compared to traditional methods such as logistic regression. The prognostic models obtained in the study for preoperative prediction of in-hospital mortality in patients referred for CABG can serve as a basis for developing systems to support medical decision-making in patients with coronary artery disease.

Adaptively Calibrated Critic Estimates for Deep Reinforcement Learning

Accurate value estimates are important for off-policy reinforcement learning. Algorithms based on temporal difference learning typically are prone to an over- or underestimation bias building up over time. In this letter, we propose a general method called Adaptively Calibrated Critics (ACC) that uses the most recent high variance but unbiased on-policy rollouts to alleviate the bias of the low variance temporal difference targets. We apply ACC to Truncated Quantile Critics [1], which is an algorithm for continuous control that allows regulation of the bias with a hyperparameter tuned per environment. The resulting algorithm adaptively adjusts the parameter during training rendering hyperparameter search unnecessary and sets a new state of the art on the OpenAI gym continuous control benchmark among all algorithms that do not tune hyperparameters for each environment. ACC further achieves improved results on different tasks from the Meta-World robot benchmark. Additionally, we demonstrate the generality of ACC by applying it to TD3 [2] and showing an improved performance also in this setting.

Hyperparameter optimization of data-driven AI models on HPC systems

In the European Center of Excellence in Exascale Computing ”Research on AI- and Simulation-Based Engineering at Exascale” (CoE RAISE), researchers develop novel, scalable AI technologies towards Exascale. This work exercises High Performance Computing resources to perform large-scale hyperparameter optimization using distributed training on multiple compute nodes. This is part of RAISE’s work on data-driven use cases which leverages AI- and HPC cross-methods developed within the project. In response to the demand for parallelizable and resource efficient hyperparameter optimization methods, advanced hyperparameter search algorithms are benchmarked and compared. The evaluated algorithms, including Random Search, Hyperband and ASHA, are tested and compared in terms of both accuracy and accuracy per compute resources spent. As an example use case, a graph neural network model known as MLPF, developed for Machine Learned Particle-Flow reconstruction, acts as the base model for optimization. Results show that hyperparameter optimization significantly increased the performance of MLPF and that this would not have been possible without access to large-scale High Performance Computing resources. It is also shown that, in the case of MLPF, the ASHA algorithm in combination with Bayesian optimization gives the largest performance increase per compute resources spent out of the investigated algorithms.

Integrated data-driven modeling and experimental optimization of granular hydrogel matrices

Integrated data-driven modeling and experimental optimization of granular hydrogel matrices

Drive System Inverter Modeling Using Symbolic Regression

For accurate and efficient control performance of electrical drives, precise values of phase voltages are required. In order to achieve control of the electric drive, the development of mathematical models of the system and its parts is often approached. Data-driven modeling using artificial intelligence can often be unprofitable due to the large amount of computing resources required. To overcome this problem, the idea is to investigate if a genetic programming–symbolic regressor (GPSR) algorithm could be used to obtain simple symbolic expressions which could estimate the mean phase voltages (black-box inverter model) and duty cycles (black-box compensation scheme) with high accuracy using a publicly available dataset. To obtain the best symbolic expressions using GPSR, a random hyperparameter search method and 5-fold cross-validation were developed. The best symbolic expressions were chosen based on their estimation performance, which was measured using the coefficient of determination (R2), mean absolute error (MAE), and root mean squared error (RMSE). The best symbolic expressions for the estimation of mean phase voltages achieved R2, MAE, and RMSE values of 0.999, 2.5, and 2.8, respectively. The best symbolic expressions for the estimation of duty cycles achieved R2, MAE, and RMSE values of 0.9999, 0.0027, and 0.003, respectively. The originality of this work lies in the application of the GPSR algorithm, which, based on a mathematical equation it generates, can estimate the value of mean phase voltages and duty cycles in a three-phase inverter. Using the obtained model, it is possible to estimate the given aforementioned values. Such high-performing estimation represents an opportunity to replace expensive online equipment with a cheaper, more precise, and faster approach, such as a GPSR-based model. The presented procedure shows that the symbolic expression for the accurate estimation of mean phase voltages and duty cycles can be obtained using the GPSR algorithm.

Classification of Wall Following Robot Movements Using Genetic Programming Symbolic Classifier

The navigation of mobile robots throughout the surrounding environment without collisions is one of the mandatory behaviors in the field of mobile robotics. The movement of the robot through its surrounding environment is achieved using sensors and a control system. The application of artificial intelligence could potentially predict the possible movement of a mobile robot if a robot encounters potential obstacles. The data used in this paper is obtained from a wall-following robot that navigates through the room following the wall in a clockwise direction with the use of 24 ultrasound sensors. The idea of this paper is to apply genetic programming symbolic classifier (GPSC) with random hyperparameter search and 5-fold cross-validation to investigate if these methods could classify the movement in the correct category (move forward, slight right turn, sharp right turn, and slight left turn) with high accuracy. Since the original dataset is imbalanced, oversampling methods (ADASYN, SMOTE, and BorderlineSMOTE) were applied to achieve the balance between class samples. These over-sampled dataset variations were used to train the GPSC algorithm with a random hyperparameter search and 5-fold cross-validation. The mean and standard deviation of accuracy (ACC), the area under the receiver operating characteristic (AUC), precision, recall, and F1−score values were used to measure the classification performance of the obtained symbolic expressions. The investigation showed that the best symbolic expressions were obtained on a dataset balanced with the BorderlineSMOTE method with ACC¯±SD(ACC), AUC¯macro±SD(AUC), Precision¯macro±SD(Precision), Recall¯macro±SD(Recall), and F1−score¯macro±SD(F1−score) equal to 0.975×1.81×10−3, 0.997±6.37×10−4, 0.975±1.82×10−3, 0.976±1.59×10−3, and 0.9785±1.74×10−3, respectively. The final test was to use the set of best symbolic expressions and apply them to the original dataset. In this case the ACC¯±SD(ACC), AUC¯±SD(AUC), Precision¯±SD(Precision), Recall¯±SD(Recall), and F1−score¯±SD(F1−Score) are equal to 0.956±0.05, 0.9536±0.057, 0.9507±0.0275, 0.9809±0.01, 0.9698±0.00725, respectively. The results of the investigation showed that this simple, non-linearly separable classification task could be solved using the GPSC algorithm with high accuracy.

Self-Attention Mechanism-Based Multi-Channel QoT Estimation in Optical Networks

It is essential to estimate the quality of transmission (QoT) of lightpaths before their establishment for efficient planning and operation of optical networks. Due to the nonlinear effect of fibers, the deployed lightpaths influence the QoT of each other; thus, multi-channel QoT estimation is necessary, which provides complete QoT information for network optimization. Moreover, the different interfering channels have different effects on the channel under test. However, the existing artificial-neural-network-based multi-channel QoT estimators (ANN-QoT-E) neglect the different effects of the interfering channels in their input layer, which affects their estimation accuracy severely. In this paper, we propose a self-attention mechanism-based multi-channel QoT estimator (SA-QoT-E) to improve the estimation accuracy of the ANN-QoT-E. In the SA-QoT-E, the input features are designed as a sequence of feature vectors of channels that route the same path, and the self-attention mechanism dynamically assigns weights to the feature vectors of interfering channels according to their effects on the channel under test. Moreover, a hyperparameter search method is used to optimize the SA-QoT-E. The simulation results show that, compared with the ANN-QoT-E, our proposed SA-QoT-E achieves higher estimation accuracy, and can be directly applied to the network wavelength expansion scenarios without retraining.

EasyPheno: An easy-to-use and easy-to-extend Python framework for phenotype prediction using Bayesian optimization.

Predicting complex traits from genotypic information is a major challenge in various biological domains. With easyPheno, we present a comprehensive Python framework enabling the rigorous training, comparison and analysis of phenotype predictions for a variety of different models, ranging from common genomic selection approaches over classical machine learning and modern deep learning-based techniques. Our framework is easy-to-use, also for non-programming-experts, and includes an automatic hyperparameter search using state-of-the-art Bayesian optimization. Moreover, easyPheno provides various benefits for bioinformaticians developing new prediction models. easyPheno enables to quickly integrate novel models and functionalities in a reliable framework and to benchmark against various integrated prediction models in a comparable setup. In addition, the framework allows the assessment of newly developed prediction models under pre-defined settings using simulated data. We provide a detailed documentation with various hands-on tutorials and videos explaining the usage of easyPheno to novice users. easyPheno is publicly available at https://github.com/grimmlab/easyPheno and can be easily installed as Python package via https://pypi.org/project/easypheno/ or using Docker. A comprehensive documentation including various tutorials complemented with videos can be found at https://easypheno.readthedocs.io/. Supplementary data are available at Bioinformatics Advances online.

Crop Yield Improvement with Weeds, Pest and Disease Detection

Detecting weeds, diseases, and pests in the agricultural field is a crucial concern for farmers. Weeds compete for light, nutrients, water, space and crop damage is caused by pests such as insects, mice, and birds and crop diseases pose a significant risk to food security. Farmers have relied on traditional methods to boost yield production in the past. The advancement of deep learning approaches helps to classify various types of photos in practice. For plant leaf disease detection, pest detection, and weed detection, we acquired many images of plant disease pairs, pest and weed images from varied crops. The data augmentation process is carried out because Deep Learning works effectively with larger data sets. The neural model was created using multiple types of DCNN architecture, and the models were interpreted based on their accuracy and performance. By fine-tuning the hyperparameters and layers of the DenseNet201, Mobilenet, VGG16, Hyperparameter Search and InceptionV3 on agricultural image data. The fine-tuned InceptionV3 model outperformed by 87.85% when comparing the final value. Mobilenet and VGG16, on the other hand, achieved accuracy of 91.85% and 78.71% respectively, DenseNet model performs well with 99.62% accuracy whereas the Hyperparameter Search with 71.07% accuracy. The model's effective performance in weed,plant disease and pest disease classification is described by the produced resultant value.

A Novel Extreme Value Theory Based approach to Hyperparameter Optimization

Finding the optimal values of hyperparameters is crucial for many machine learning algorithms. Algorithms like deep learning, XGBoost, etc. require us to choose among a large set of hyperparameters, thus increasing the computational requirements for a successful hyperparameter search. Another factor complicating the hyperparameter search process is the variability in performance due to choice of different initial random seeds. This variability might be caused by different initial weight assignments or by different environment interactions in the training process. In this paper an extreme value theory (EVT) based approach for hyperparameter optimization in this noisy setting is proposed. To the best of our knowledge this is the first work based on EVT for hyperparameter optimization. It is demonstrated through extensive experimentation on many real world datasets (including a medical diabetes dataset consisting of several medical predictor variables), that by using only a fraction of observations near optimal hyperparameter values can be obtained using the parameters of a generalized extreme value distribution for the test set performance metric of interest in the algorithm.

Predicting Individual Patient Platelet Demand in a Large Tertiary Care Hospital Using Machine Learning

Introduction: An increasing shortage of donor blood is expected, considering the demographic change in Germany. Due to the short shelf life and varying daily fluctuations in consumption, the storage of platelet concentrates (PCs) becomes challenging. This emphasizes the need for reliable prediction of needed PCs for the blood bank inventories. Therefore, the objective of this study was to evaluate multimodal data from multiple source systems within a hospital to predict the number of platelet transfusions in 3 days on a per-patient level. Methods: Data were collected from 25,190 (42% female and 58% male) patients between 2017 and 2021. For each patient, the number of received PCs, platelet count blood tests, drugs causing thrombocytopenia, acute platelet diseases, procedures, age, gender, and the period of a patient’s hospital stay were collected. Two models were trained on samples using a sliding window of 7 days as input and a day 3 target. The model predicts whether a patient will be transfused 3 days in the future. The model was trained with an excessive hyperparameter search using patient-level repeated 5-fold cross-validation to optimize the average macro F2-score. Results: The trained models were tested on 5,022 unique patients. The best-performing model has a specificity of 0.99, a sensitivity of 0.37, an area under the precision-recall curve score of 0.45, an MCC score of 0.43, and an F1-score of 0.43. However, the model does not generalize well for cases when the need for a platelet transfusion is recognized. Conclusion: A patient AI-based platelet forecast could improve logistics management and reduce blood product waste. In this study, we build the first model to predict patient individual platelet demand. To the best of our knowledge, we are the first to introduce this approach. Our model predicts the need for platelet units for 3 days in the future. While sensitivity underperforms, specificity performs reliably. The model may be of clinical use as a pretest for potential patients needing a platelet transfusion within the next 3 days. As sensitivity needs to be improved, further studies should introduce deep learning and wider patient characterization to the methodological multimodal, multisource data approach. Furthermore, a hospital-wide consumption of PCs could be derived from individual predictions.