Machine Selection Research Articles

Employee Turnover Prediction based on Particle Swarm Optimization - Support Vector Machine

Employee turnover is getting more and more attention in the human resources field. Unexpected turnover of employees is blamed for the loss of work handover. As a result, predicting whether employees would leave has become a crucial problem. This research aims to exploit a method combining particle swarm optimization with a support vector machine to address the employee departure prediction problem. In this study, the particle swarm optimization algorithm is used to optimize the parameter selection of the support vector machine to improve the performance of the latter. Moreover, employee information of a dataset is subject to correlation analysis before being transformed into standardization form to accelerate convergence and improve the accuracy of the support vector machine. Eventually, the support vector machine combined with particle swarm optimization is of best performance in accuracy score, precision score and F1 score, respectively reaching 0.873, 0.947 and 0.784. In conclusion, this method addresses the employee turnover prediction problem effectively which also provides a new direction for applying swarm intelligence algorithms.

Dynamic Integrated Scheduling of Production Equipment and Automated Guided Vehicles in a Flexible Job Shop Based on Deep Reinforcement Learning

The high-quality development of the manufacturing industry necessitates accelerating its transformation towards high-end, intelligent, and green development. Considering logistics resource constraints, the impact of dynamic disturbance events on production, and the need for energy-efficient production, the integrated scheduling of production equipment and automated guided vehicles (AGVs) in a flexible job shop environment is investigated in this study. Firstly, a static model for the integrated scheduling of production equipment and AGVs (ISPEA) is developed based on mixed-integer programming, which aims to optimize the maximum completion time and total production energy consumption (EC). In recent years, reinforcement learning, including deep reinforcement learning (DRL), has demonstrated significant advantages in handling workshop scheduling issues with sequential decision-making characteristics, which can fully utilize the vast quantity of historical data accumulated in the workshop and adjust production plans in a timely manner based on changes in production conditions and demand. Accordingly, a DRL-based approach is introduced to address the common production disturbances in emergency order insertions. Combined with the characteristics of the ISPEA problem and an event-driven strategy for handling dynamic events, four types of agents, namely workpiece selection, machine selection, AGV selection, and target selection agents, are set up, which refine workshop production status characteristics as observation inputs and generate rules for selecting workpieces, machines, AGVs, and targets. These agents are trained offline using the QMIX multi-agent reinforcement learning framework, and the trained agents are utilized to solve the dynamic ISPEA problem. Finally, the effectiveness of the proposed model and method is validated through a comparison of the solution performance with other typical optimization algorithms for various cases.

Machine Learning Assisted Stroke Prediction in Mechanical Circulatory Support: Predictive Role of Systemic Mitochondrial Dysfunction.

Stroke continues to be a major adverse event in advanced congestive heart failure (CHF) patients after continuous-flow left ventricular assist device (CF-LVAD) implantation. Abnormalities in mitochondrial oxidative phosphorylation (OxPhos) have been critically implicated in the pathogenesis of neurodegenerative diseases and cerebral ischemia. We hypothesize that prior stroke may be associated with systemic mitochondrial OxPhos abnormalities, and impaired more in post-CF-LVAD patients with risk of developing new stroke. We studied 50 CF-LVAD patients (25 with prior stroke, 25 without); OxPhos complex proteins (complex I [C.I]-complex V [C.V]) were measured in blood leukocytes. Both at baseline (pre-CF-LVAD) and postoperatively (post-CF-LVAD), the prior-stroke group had significantly lower C.I, complex II (C.II), complex IV (C.IV), and C.V proteins when compared to the no-prior-stroke group. Oxidative phosphorylation proteins were significantly decreased in prior-stroke group at post-CF-LVAD compared to pre-CF-LVAD. Machine learning Least Absolute Shrinkage and Selection Operator (LASSO) and Random Forest modeling identified six prognostic factors that predicted postoperative stroke with an area under the receiver operating characteristic (ROC) curve (AUC) of 0.93. Oxidative phosphorylation protein reduction appeared to be associated with the new stroke after implantation. Our study found for the first time the existence of mitochondrial dysfunction at the peripheral level in CHF patients with prior ischemic stroke even before CF-LVAD implantation. The changes in OxPhos protein expression could serve as biomarkers in predicting new post-CF-LVAD strokes.

Mechanical Properties of Spunlace Non-Wovens with a Porous Structure

The paper describes the influence of the drum system construction of two modern carding machines on the porous structure of spunlace non-wovens composed of polyester and viscose. The non-woven fabric structure, including the number and size of the pores, determines the tensile strength of the composites obtained. The spunlace non-wovens were subjected to tensile strength tests in the machine, and cross-directions and microscopic observations of their structure were made. The results of the experiments were used to determine the relationship between the strength of the material and the porosity of its structure. This knowledge was used to prepare recommendations for the manufacturer of wet wipes in order to enable the selection of a carding machine for the mass production of final products with strength properties that meet market requirements and satisfy the end customer.

Identification of novel diagnostic biomarkers associated with liver metastasis in colon adenocarcinoma by machine learning

BackgroundLiver metastasis is one of the primary causes of poor prognosis in colon adenocarcinoma (COAD) patients, but there are few studies on its biomarkers.MethodsThe Cancer Genome Atlas (TCGA)-COAD, GSE41258, and GSE49355 datasets were acquired from the public database. Differentially expressed genes (DEGs) between liver metastasis and primary tumor samples in COAD were identified by limma, and functional enrichment analysis were performed. MuTect2 and maftools were used to measure somatic mutation rates, while ADTEx was used to measure copy number variations (CNVs). The intersection of three machine learning methods, support vector machine (SVM), Random Forest, and least absolute shrinkage and selection operator (LASSO), is utilized to screen biomarkers, and their diagnostic performance is subsequently validated. The correlation between biomarkers and immune cells infiltration was analyzed by Spearman method.Results47 DEGs between liver metastasis and primary tumor samples in COAD were obtained, which were mainly enriched in the complement and coagulation, extracellular matrix (ECM), and peptidase regulator activity, etc. 38 out of 47 DEGs had mutations and exhibited a high frequency of CNV amplification or deletion. Furthermore, 3 biomarkers (MMP3, MAB21L2, and COLEC11) were screened, which showed good diagnostic performance. The proportion of multiple immune cells, such as B cells naive, T cells CD4 naive, Monocytes, and Dendritic cells resting, was higher in liver metastasis samples than that in primary tumor samples. Meanwhile, MMP3, MAB21L2, and COLEC11 exhibited an outstanding correlation with immune cells infiltration.ConclusionIn short, 3 biomarkers with good diagnostic efficacy were identified, providing a new perspective of therapeutic targets for liver metastasis in COAD.

A bi-population cooperative scatter search algorithm for distributed hybrid flow shop scheduling with machine breakdown

The occurrence of machine breakdowns is a frequent and dynamic phenomenon in the production process. The implementation of effective preventive measures can mitigate such events and result in reduced production costs. This paper investigates the distributed hybrid flow shop scheduling problem with machine breakdown (DHFSSPB) considering short maintenance time. The bi-population cooperative scatter search (BCSS) algorithm is proposed to address the DHFSSPB, wherein the search for the optimal scheduling sequence is transformed into genetic evolution aiming to obtain a gene chain with both minimum lower bound and minimum cost attributes. Firstly, the DHFSSPB problem is modeled through a combination of predictive maintenance strategy and right-shift rescheduling rule. Subsequently, a diversification approach is developed to facilitate attribute inheritance, enhance the efficiency of job allocation, and establish a reference set. The reference set is partitioned into two subpopulations based on lower bound attributes and cost attributes, respectively. The corresponding hybrid search strategies are designed to enhance the efficiency of job sorting and machine selection for subpopulations with distinct attributes. The cooperative evolution between subpopulations occurs through the competitive interaction and fusion of individuals. An enhanced reinforcement learning approach is proposed to expedite the acceleration of individual attribute evolution by leveraging evolutionary knowledge acquired from populations, thereby effectively guiding the evolutionary trajectories of individuals. Additionally, a method for evaluating the population during the learning process is developed based on problem characteristics to enhance learning efficiency. Experimental results demonstrate that BCSS outperforms the comparative algorithm in solving the DHFSSPB.

Impacts of Material and Machine on the Variation of Additively Manufactured Cooling Channels

Abstract While additive manufacturing (AM) can reduce component development time and create unique internal cooling designs, the AM process also introduces several sources of variability, such as the selection of machine, material, and print parameters. Because of these sources, wide variations in a part's geometrical accuracy and surface roughness levels can occur, especially for small internal cooling features that are difficult to post-process. This study investigates how the selection of machine and material in the AM process influences variations in surface quality and deviations from the design intent. Two microscale cooling geometries were tested: wavy channels and diamond-shaped pin fins. Test coupons were fabricated with five different additive machines and four materials using process parameters recommended by the manufacturers. The as-built geometry was measured non-destructively with computed tomography scans. To evaluate surface roughness, the coupons were cut open and examined using a laser microscope. Three distinct roughness profiles on the coupon surfaces were captured including upskin, downskin, and channel walls built at 90 deg to the build plate. Results indicated that both material and machine contribute to producing different roughness levels and very different surface morphologies. The roughness levels on the downskin surfaces are significantly greater than on the upskin or sidewall surfaces. Geometric analysis revealed that while the hydraulic diameter of all coupons was well captured, the pin cross section varied considerably. Along with characterizing the coupon surfaces, cooling performance was investigated by experimentally measuring friction factor and heat transfer. The variations in surface morphology as a function of material and machine resulted in heat transfer fluctuating by up to 50% between coupons featuring wavy channels and 26% for coupons with pin fin arrays. Increased arithmetic mean surface roughness led to increased heat transfer and pressure drop; however, a secondary driver in the performance of the wavy channels was found to be the roughness morphology, which could be described using the surface skewness and kurtosis.

Construction and validation of a nomogram model for lymph node metastasis of stage II-III gastric cancer based on machine learning algorithms.

Gastric cancer, a pervasive malignancy globally, often presents with regional lymph node metastasis (LNM), profoundly impacting prognosis and treatment options. Existing clinical methods for determining the presence of LNM are not precise enough, necessitating the development of an accurate risk prediction model. Our primary objective was to employ machine learning algorithms to identify risk factors for LNM and establish a precise prediction model for stage II-III gastric cancer. A study was conducted at Renji Hospital Affiliated to Shanghai Jiao Tong University School of Medicine between May 2010 and December 2022. This retrospective study analyzed 1147 surgeries for gastric cancer and explored the clinicopathological differences between LNM and non-LNM cohorts. Utilizing univariate logistic regression and two machine learning methodologies-Least absolute shrinkage and selection operator (LASSO) and random forest (RF)-we identified vascular invasion, maximum tumor diameter, percentage of monocytes, hematocrit (HCT), and lymphocyte-monocyte ratio (LMR) as salient factors and consolidated them into a nomogram model. The area under the receiver operating characteristic (ROC) curve (AUC), calibration curves, and decision curves were used to evaluate the test efficacy of the nomogram. Shapley Additive Explanation (SHAP) values were utilized to illustrate the predictive impact of each feature on the model's output. Significant differences in tumor characteristics were discerned between LNM and non-LNM cohorts through appropriate statistical methods. A nomogram, incorporating vascular invasion, maximum tumor diameter, percentage of monocytes, HCT, and LMR, was developed and exhibited satisfactory predictive capabilities with an AUC of 0.787 (95% CI: 0.749-0.824) in the training set and 0.753 (95% CI: 0.694-0.812) in the validation set. Calibration curves and decision curves affirmed the nomogram's predictive accuracy. In conclusion, leveraging machine learning algorithms, we devised a nomogram for precise LNM risk prognostication in stage II-III gastric cancer, offering a valuable tool for tailored risk assessment in clinical decision-making.

IDE-SVM algorithm and it's usage in physical layer security method for IoT

To address the problem that there is no standard for the parameter selection of support vector machine (SVM) algorithm, a parameter optimization selection method based on the integrated improved differential evolution (IDE) algorithm is proposed, which uses the minimization of the classification error rate as the optimization criterion and the improved differential evolution algorithm to optimize the combination of SVM parameters to obtain an SVM algorithm with higher classification accuracy. At the same time, to avoid the basic DE algorithm from falling into local optimum when solving the SVM parameter selection problem and to improve the search efficiency of the DE algorithm. In this paper, an improved differential evolutionary algorithm is proposed to obtain a DE algorithm with faster convergence and higher accuracy by using the circular arc function for adaptive control of the variance scaling factor F and the crossover probability factor R, combining with the random newborn individual replacement operation. Based on this, an IDE-SVM IoT physical layer security method based on IDE-SVM is proposed. The experimental results show that the authentication accuracy of the physical layer security method based on the improved SVM algorithm is higher than others.

Study on the forging process of artificial bone plates of SUS 316L

Abstract SUS 316L stainless steel, valued for its strength and biocompatibility, is a popular choice for artificial bone plates. Forged SUS 316L plates offer advantages like enhanced strength, improved fatigue resistance, reduced material waste, and increased production efficiency. Consequently, the forging technology and die design for SUS 316L is crucial in the manufacturing of high quality artificial bone plates. Finite element analysis simulates the forging process of SUS 316L bone plates, guiding the selection of the appropriate forging machine, die design, and process parameters based on the resulting forging load, stress distribution, and filling condition. Ultimately, this approach facilitates the creation of high-quality SUS 316L bone plates with precise dimensional accuracy.

Construction and evaluation of a liver cancer risk prediction model based on machine learning.

Liver cancer is one of the most prevalent malignant tumors worldwide, and its early detection and treatment are crucial for enhancing patient survival rates and quality of life. However, the early symptoms of liver cancer are often not obvious, resulting in a late-stage diagnosis in many patients, which significantly reduces the effectiveness of treatment. Developing a highly targeted, widely applicable, and practical risk prediction model for liver cancer is crucial for enhancing the early diagnosis and long-term survival rates among affected individuals. To develop a liver cancer risk prediction model by employing machine learning techniques, and subsequently assess its performance. In this study, a total of 550 patients were enrolled, with 190 hepatocellular carcinoma (HCC) and 195 cirrhosis patients serving as the training cohort, and 83 HCC and 82 cirrhosis patients forming the validation cohort. Logistic regression (LR), support vector machine (SVM), random forest (RF), and least absolute shrinkage and selection operator (LASSO) regression models were developed in the training cohort. Model performance was assessed in the validation cohort. Additionally, this study conducted a comparative evaluation of the diagnostic efficacy between the ASAP model and the model developed in this study using receiver operating characteristic curve, calibration curve, and decision curve analysis (DCA) to determine the optimal predictive model for assessing liver cancer risk. Six variables including age, white blood cell, red blood cell, platelet counts, alpha-fetoprotein and protein induced by vitamin K absence or antagonist II levels were used to develop LR, SVM, RF, and LASSO regression models. The RF model exhibited superior discrimination, and the area under curve of the training and validation sets was 0.969 and 0.858, respectively. These values significantly surpassed those of the LR (0.850 and 0.827), SVM (0.860 and 0.803), LASSO regression (0.845 and 0.831), and ASAP (0.866 and 0.813) models. Furthermore, calibration and DCA indicated that the RF model exhibited robust calibration and clinical validity. The RF model demonstrated excellent prediction capabilities for HCC and can facilitate early diagnosis of HCC in clinical practice.

Machine Learning-Based Feature Extraction and Selection

Machine Learning-Based Feature Extraction and Selection

Molecular insights into programmed cell death in esophageal squamous cell carcinoma.

Esophageal squamous cell carcinoma (ESCC) is a deadly type of esophageal cancer. Programmed cell death (PCD) is an important pathway of cellular self-extermination and is closely involved in cancer progression. A detailed study of its mechanism may contribute to ESCC treatment. We obtained expression profiling data of ESCC patients from public databases and genes related to 12 types of PCD from previous studies. Hub genes in ESCC were screened from PCD-related genes applying differential expression analysis, machine learning analysis, linear support vector machine (SVM), random forest and Least Absolute Shrinkage and Selection Operator (LASSO) regression analysis. In addition, based on the HTFtarget and TargetScan databases, transcription factors (TFs) and miRNAs interacting with the hub genes were selected. The relationship between hub genes and immune cells were analyzed using the CIBERSORT algorithm. Finally, to verify the potential impact of the screened hub genes on ESCC occurrence and development, a series of in vitro cell experiments were conducted. We screened 149 PCD-related DEGs, of which five DEGs (INHBA, LRRK2, HSP90AA1, HSPB8, and EIF2AK2) were identified as the hub genes of ESCC. The area under the curve (AUC) of receiver operating characteristic (ROC) curve of the integrated model developed using the hub genes reached 0.997, showing a noticeably high diagnostic accuracy. The number of TFs and miRNAs regulating hub genes was 105 and 22, respectively. INHBA, HSP90AA1 and EIF2AK2 were overexpressed in cancer tissues and cells of ESCC. Notably, INHBA knockdown suppressed ECSS cell migration and invasion and altered the expression of important apoptotic and survival proteins. This study identified significant molecules with promising accuracy for the diagnosis of ESCC, which may provide a new perspective and experimental basis for ESCC research.

Identification of necroptosis-related genes in ankylosing spondylitis by bioinformatics and experimental validation.

The pathogenesis of ankylosing spondylitis (AS) remains unclear, and while recent studies have implicated necroptosis in various autoimmune diseases, an investigation of its relationship with AS has not been reported. In this study, we utilized the Gene Expression Omnibus database to compare gene expressions between AS patients and healthy controls, identifying 18 differentially expressed necroptosis-related genes (DENRGs), with 8 upregulated and 10 downregulated. Through the application of three machine learning algorithms-least absolute shrinkage and selection operation, support vector machine-recursive feature elimination and random forest-two hub genes, FASLG and TARDBP, were pinpointed. These genes demonstrated high specificity and sensitivity for AS diagnosis, as evidenced by receiver operating characteristic curve analysis. These findings were further supported by external datasets and cellular experiments, which confirmed the downregulation of FASLG and upregulation of TARDBP in AS patients. Immune cell infiltration analysis suggested that CD4+ T cells, CD8+ T cells, NK cells and neutrophils may be associated with the development of AS. Notably, in the group with high FASLG expression, there was a significant infiltration of CD8+ T cells, memory-activated CD4+ T cells and resting NK cells, with relatively less infiltration of memory-resting CD4+ T cells and neutrophils. Conversely, in the group with high TARDBP expression, there was enhanced infiltration of naïve CD4+ T cells and M0 macrophages, with a reduced presence of memory-resting CD4+ T cells. In summary, FASLG and TARDBP may contribute to AS pathogenesis by regulating the immune microenvironment and immune-related signalling pathways. These findings offer new insights into the molecular mechanisms of AS and suggest potential new targets for therapeutic strategies.

Cross-Validation Visualized: A Narrative Guide to Advanced Methods

This study delves into the multifaceted nature of cross-validation (CV) techniques in machine learning model evaluation and selection, underscoring the challenge of choosing the most appropriate method due to the plethora of available variants. It aims to clarify and standardize terminology such as sets, groups, folds, and samples pivotal in the CV domain, and introduces an exhaustive compilation of advanced CV methods like leave-one-out, leave-p-out, Monte Carlo, grouped, stratified, and time-split CV within a hold-out CV framework. Through graphical representations, the paper enhances the comprehension of these methodologies, facilitating more informed decision making for practitioners. It further explores the synergy between different CV strategies and advocates for a unified approach to reporting model performance by consolidating essential metrics. The paper culminates in a comprehensive overview of the CV techniques discussed, illustrated with practical examples, offering valuable insights for both novice and experienced researchers in the field.

A multidimensional probabilistic model based evolutionary algorithm for the energy-efficient distributed flexible job-shop scheduling problem

With escalating environmental effects, the spotlight on low-carbon manufacturing has garnered significant attention. The rise of distributed production has emerged as a prominent trend in response to the imperatives of economic globalization. This article focuses on addressing the energy-efficient distributed flexible job-shop scheduling problem (EE_DFJSP), with the aim of minimizing both makespan and total energy consumption (TEC) simultaneously. The production process contains four pivotal phases: 1) job assignment in distributed factories; 2) machine selection within factories; 3) operation allocation on flexible machines; and 4) machine speed adjustment for processing. Given the problem's multi-phase and strong coupling characteristics, it is imperative to develop a promising evolutionary algorithm (EA) for EE_DFJSP. To tackle this challenge, we propose a multidimensional probabilistic model-based EA (MPMEA) paradigm. First, problem-specific encoding and decoding schemes are developed based on the solution features of EE_DFJSP. Second, a hybrid initialization strategy incorporating four heuristic rules is devised to yield an initial population with diversity. Third, an effective union probabilistic model (UPM) is formulated to learn promising patterns from superior solutions, and an efficient sampling strategy is designed to produce high-quality offspring individuals. To achieve a balance between global exploration and local exploitation, problem-specific multiple neighborhood operators are proposed to perform an in-depth local search. Furthermore, a two-stage energy-saving speed adjustment strategy is designed for the superior solutions obtained through local search. Finally, computational comparisons and simulation studies are conducted to validate the effectiveness and superiority of the MPMEA in effectively addressing EE_DFJSP.

SCRUB: a novel energy-efficient virtual machines selection and migration scheme in cloud data centers

SCRUB: a novel energy-efficient virtual machines selection and migration scheme in cloud data centers

Novel Technical Parameters-Based Classification of Harvesters Using Principal Component Analysis and Q-Type Cluster Model

The advancement of agricultural mechanization necessitates precise and standardized classification based on technical characteristics to enhance green, efficient, and high-quality development. The current lack of scientific and standardized definitions and classifications for various types of agricultural machinery has become a bottleneck, complicating the machine selection and affecting the compatibility of the machinery with optimized field operations. To address this complexity, we propose a comprehensive classification method that integrates principal component analysis (PCA), cluster analysis, and the qualitative analysis of the functional components for defining and scientifically classifying harvesters. The key functional and technical properties of harvesters were analyzed, and eight primary parameters (machine weight, cutting width, feed rate, rated power, overall machine length, width, height, and working efficiency) were selected, supplemented by nine key functional components (walking mechanism, cutting device, threshing device, separating device, cleaning device, grain collecting device, grain unloading device, cabin, and track size). In the first step, principal component analysis was performed to reduce the dimensionality of the parameters, yielding three principal components with contribution rates of 41.610%, 28.579%, and 15.134%, respectively. One primary parameter from each component was selected for further analysis. In the second stage, Q-type cluster analysis classified the harvesters based on the squared Euclidean distance between the operational parameters, resulting in three classes of harvesters. Finally, functional component analysis provided detailed insights, further refining the classification into four major categories: mini, small, medium, and large harvesters. The results of this work provide a scientific basis for the definition and classification of the harvester products available on the market. This method offers a robust framework for the rational selection and planning of agricultural machinery, promoting sustainable mechanization with a focus on technical parameters and functional attributes.

Peningkatan Daya Saing UMKM Mimi Roti melalui Penerapan Teknologi Pengemasan dan Pemasaran Digital

This time, the community service program partner is Mimi Roti Micro, Small and Medium Enterprises (MSMEs) in Watugede, Singosari, Malang Regency. Founded in 1999, Mimi Roti MSME has experienced rapid development by reaching a production capacity of 100 donuts and burgers per day. However, Mrs. Alfiyah, as the owner of UMKM Mimi Roti, experienced significant obstacles in terms of product packaging. Using simple plastic clips causes the bread's shelf life to be only three days, limiting the growth potential and competitiveness of Mimi Roti MSMEs in the market. This service activity aims to help Mimi Roti MSMEs overcome packaging problems and increase their competitiveness in the market. To overcome this problem, the community service team took several steps, namely: analysis of partner needs, research on continuous band sealer (CBS) machines, selection and procurement of CBS machines, packaging redesign, training on using CBS machines, training on digital marketing strategies, monitoring, and evaluation. Applying the CBS machine and packaging redesign increased the shelf life of bread to 7 days, improved the quality and attractiveness of the packaging, and increased sales by up to 20%. Community service activities at MSME Mimi Roti have succeeded in helping MSME Mimi Roti overcome packaging problems, increase competitiveness, and expand its market reach. Implementing better packaging technology, attractive packaging redesign, and effective digital marketing strategies have proven to positively impact product quality, sales, and profits of Mimi Roti MSMEs.

Abstract PO4-07-01: Deep learning guided radiation therapy machine selection via a multi-headed dose prediction model

Abstract Background: Accelerated Partial Breast Irradiation (APBI) is gaining popularity for the treatment of appropriately selected breast cancer patients in lieu of whole breast irradiation regimens due to its comparable clinical outcomes and improved patient convenience and quality-of-life. Currently, several machines for delivery of APBI are available such as CyberKnife, Elekta Unity, Varian Ethos and Varian TrueBeam. Each treatment machine yields a unique distribution of radiation dose, so a given modality may afford particular advantages or liabilities based on a patient’s anatomy. Existing methods for identification of an ideal modality that best suits a patient’s unique anatomical features are generally lacking, manual, and resource intensive. To overcome this limitation and personalize modality selection, we propose a deep learning (DL) based multi-modality dose prediction model. The model takes a patient’s computed tomography (CT) scan, segmented planning target volumes (PTVs) and segmented organs at risk (OARs) as input, and outputs predicted dose distributions for all trained modalities. Differences may then be evaluated between each radiation treatment modality’s corresponding dose distribution and dose volumetric histograms (DVH) to select the optimal modality for each patient. Methods: Our dataset contains 16 partial breast patients, each with 4 unique treatment plans generated for 4 distinct modalities – CyberKnife, Elekta Unity, Varian Ethos and Varian Truebeam. 10, 2, and 4 patients were dedicated to the train, validation and test datasets respectively. Our model’s architecture is a U-net with a shared encoder and 4 independent dose decoders, in which each decoder predicts dose for one modality. Each training patient’s CT scan, PTVs, OARs, and a PTV distance map are input to the model during training, and the shared encoder shares the input data’s latent representations with each of the 4 decoders. The 4 distinct dose distributions from each training patient are used as labels for the 4 corresponding decoders to supervise model training. The mean squared error, mean squared log error, and DVH losses are minimized during model training. The model trained until validation loss converged. The 4 dose distributions were predicted in ~ 1 minute per test patient, yielding 16 dose distributions in the test dataset. Model performance across the test dataset was evaluated by calculation of the mean absolute percent error (MAPE) of dose differences within the patient’s body between the model’s predicted doses and label doses, relative to patient prescription dose. Results: The performance of the multi-modal dose prediction model was evaluated in terms of dose differences throughout each patient’s body across the 16 test dose distributions as compared to the actual planned doses. The average MAPE of the multi-modal dose predictions was 1.27 ± 0.38% of a patient’s prescription dose as compared to the actual planned dose across the test dataset. For a prescription dose of 3000 cGy, that corresponds to an average MAPE of 38.1 ± 11.4 cGy. Conclusion: Our multi-modality dose prediction model’s novelty lies in its rapid provision and comparison of accurate and detailed dosimetric data for each treatment modality. This model can empower treating radiation oncologists to make informed decisions in radiation therapy machine selection on a patient-by-patient basis, potentially optimizing radiation treatment plans by reduction of delivered dose to organs at risk. Table. Citation Format: Austen Maniscalco, Ezek Mathew, David Parsons, Asal Rahimi, Mona Arbab, Prasanna Alluri, Xingzhe Li, Mu-Han Lin, Steve Jiang, Dan Nguyen. Deep learning guided radiation therapy machine selection via a multi-headed dose prediction model [abstract]. In: Proceedings of the 2023 San Antonio Breast Cancer Symposium; 2023 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2024;84(9 Suppl):Abstract nr PO4-07-01.