Skip to main content

Milan Toma

New York Institute of Technology, College of Osteopathic Medicine, Faculty Member

Followers

0

Following

0

Public Views

Currently serving an Associate Professor of Clinical Sciences at the New York Institute of Technology College of Osteopathic Medicine (NYITCOM), where I am affiliated with the Department of Osteopathic Manipulative Medicine. In this role, I'm applying my knowledge of biomedical engineering to address challenges in medicine and biology, delivering innovative engineering solutions.

less

Interests

Uploads

Papers by Milan Toma

Training clinically relevant EEG–based pain detection models: a pilot machine learning study

Academia Biology, Jul 7, 2025

Assessing pain in non-communicative patients remains a critical challenge in clinical settings. T... more Assessing pain in non-communicative patients remains a critical challenge in clinical settings. This pilot study introduces a machine learning (ML) framework for pain detection using electroencephalogram (EEG) signals, leveraging methodological rigor to ensure model robustness and generalizability. EEG data were collected from a single healthy subject exposed to controlled mild pain stimuli (self-administered forearm pinches). The pipeline included signal normalization, wavelet-based feature extraction (Daubechies 4 wavelet), principal component analysis (PCA) for dimensionality reduction, and SMOTEENN to address class imbalance. A logistic regression model, optimized via Bayesian hyperparameter tuning and validated with stratified 5-fold cross-validation, achieved an accuracy of 84.62% (validation) and 75.68% (test), with precision/recall balances indicating low false-positive rates (90%/80%) and moderate sensitivity (75%/66.67%). Comparative analyses with SVM, ensemble methods, and Naive Bayes highlighted logistic regression’s superior generalization, supported by learning curves demonstrating stable convergence. Despite promising proof-of-concept results, limitations include the single-subject design and artificial pain stimulus, necessitating future validation in larger cohorts and comatose populations. This work underscores the importance of transparent ML training processes and robustness evaluations for clinical translation, advocating for methodological standards that prioritize generalizability over inflated performance metrics.

Cardiovascular Health Management in Diabetic Patients with Machine-Learning-Driven Predictions and Interventions

Applied sciences, Mar 4, 2024

The advancement of machine learning in healthcare offers significant potential for enhancing dise... more The advancement of machine learning in healthcare offers significant potential for enhancing disease prediction and management. This study harnesses the PyCaret library-a Python-based machine learning toolkit-to construct and refine predictive models for diagnosing diabetes mellitus and forecasting hospital readmission rates. By analyzing a rich dataset featuring a variety of clinical and demographic variables, we endeavored to identify patients at heightened risk for diabetes complications leading to readmissions. Our methodology incorporates an evaluation of numerous machine learning algorithms, emphasizing their predictive accuracy and generalizability to improve patient care. We scrutinized the predictive strength of each model concerning crucial metrics like accuracy, precision, recall, and the area under the curve, underlining the imperative to eliminate false diagnostics in the field. Special attention is given to the use of the light gradient boosting machine classifier among other advanced modeling techniques, which emerge as particularly effective in terms of the Kappa statistic and Matthews correlation coefficient, suggesting robustness in prediction. The paper discusses the implications of diabetes management, underscoring interventions like lifestyle changes and pharmacological treatments to avert long-term complications. Through exploring the intersection of machine learning and health informatics, the study reveals pivotal insights into algorithmic predictions of diabetes readmission. It also emphasizes the necessity for further research and development to fully incorporate machine learning into modern diabetes care to prompt timely interventions and achieve better overall health outcomes. The outcome of this research is a testament to the transformative impact of automated machine learning in the realm of healthcare analytics.

Machine learning for medical image classification

Academia Medicine, Dec 23, 2024

This review article focuses on the application of machine learning (ML) algorithms in medical ima... more This review article focuses on the application of machine learning (ML) algorithms in medical image classification. It highlights the intricate process involved in selecting the most suitable ML algorithm for predicting specific medical conditions, emphasizing the critical role of real-world data in testing and validation. It navigates through various ML methods utilized in healthcare, including Supervised Learning, Unsupervised Learning, Self-Supervised Learning, Deep Neural Networks, Reinforcement Learning, and Ensemble Methods. The challenge lies not just in the selection of an ML algorithm but in identifying the most appropriate one for a specific task as well, given the vast array of options available. Each unique dataset requires a comparative analysis to determine the best-performing algorithm. However, testing all available algorithms is impractical. This article examines the performance of various ML algorithms in recent studies, focusing on their applications across different imaging modalities for diagnosing specific medical conditions. It provides a summary of these studies, offering a starting point for those seeking to select the most suitable ML algorithm for specific medical conditions and imaging modalities.

Biomechanical Effects of Seizures on Cerebral Dynamics and Brain Stress

Brain sciences, Mar 27, 2024

This article is an open access article distributed under the terms and conditions of the Creative... more

Haemodynamic changes in visceral hybrid repairs of type III and type V thoracoabdominal aortic aneurysms

Scientific Reports, Aug 22, 2023

The visceral hybrid procedure combining retrograde visceral bypass grafting and completion endova... more The visceral hybrid procedure combining retrograde visceral bypass grafting and completion endovascular stent grafting is a feasible alternative to conventional open surgical or wholly endovascular repairs of thoracoabdominal aneurysms (TAAA). However, the wide variability in visceral hybrid configurations means that a priori prediction of surgical outcome based on haemodynamic flow profiles such as velocity pattern and wall shear stress post repair remain challenging. We sought to appraise the clinical relevance of computational fluid dynamics (CFD) analyses in the setting of visceral hybrid TAAA repairs. Two patients, one with a type III and the other with a type V TAAA, underwent successful elective and emergency visceral hybrid repairs, respectively. Flow patterns and haemodynamic parameters were analysed using reconstructed pre-and post-operative CT scans. Both type III and type V TAAAs showed highly disturbed flow patterns with varying helicity values preoperatively within their respective aneurysms. Low time-averaged wall shear stress (TAWSS) and high endothelial cell action potential (ECAP) and relative residence time (RRT) associated with thrombogenic susceptibility was observed in the posterior aspect of both TAAAs preoperatively. Despite differing bypass configurations in the elective and emergency repairs, both treatment options appear to improve haemodynamic performance compared to preoperative study. However, we observed reduced TAWSS in the right iliac artery (portending a theoretical risk of future graft and possibly limb thrombosis), after the elective type III visceral hybrid repair, but not the emergency type V repair. We surmise that this difference may be attributed to the higher neo-bifurcation of the aortic stent graft in the type III as compared to the type V repair. Our results demonstrate that CFD can be used in complicated visceral hybrid repair to yield potentially actionable predictive insights with implications on surveillance and enhanced post-operative management, even in patients with complicated geometrical bypass configurations. Overall treatment for thoracoabdominal aneurysms (TAAA) has improved considerably over the past three decades because of advances in surgical techniques. Although open surgical repair remains popular, it has since been supplanted by total endovascular repair 1,2. Unfortunately, these operations still carry significant risks of complications and mortality, with the potential of ischemic insult to the spinal cord, kidneys and other visceral organs 3,4. The visceral hybrid approach 5 involves surgically rerouting the visceral and renal branches followed by endovascular exclusion of the TAAA with an endograft. This method is an appealing alternative as it avoids

Comparative Analysis of Machine Learning Models for Image Detection of Colonic Polyps versus Resected Polyps

(1) Background: Colon polyps are common protrusions in the colon’s lumen with potential risks of ... more (1) Background: Colon polyps are common protrusions in the colon’s lumen with potential risks of developing colorectal cancer. Early detection and intervention of these polyps are vital for reducing colorectal cancer incidence and mortality rates. This research aims to evaluate and compare the performance of three machine-learning image classification models’ performance in detecting and classifying colon polyps. (2) Methods: The performance of three machine learning image classification models, Google Teachable Machine (GTM), Roboflow3 (RF3), and You Only Look Once version 8 (YOLOV8), in the detection and classification of colon polyps were evaluated. The study used a dataset of colonoscopy images of normal colon, polyps, and resected polyps. The study assessed the models’ ability to correctly classify the images into their respective classes using precision, recall, and F1 score generated from confusion matrix analysis and performance graphs. (3) Results: All three models successfully distinguished between normal colon, polyps, and resected polyps in colonoscopy images. GTM achieved the highest accuracies: 0.99, with consistent precision, recall, and F1 scores of 1.00 for the `normal’ class, 0.97 - 1.00 for `polyps,’ and 0.97 - 1.00 for `resected polyps.’ While GTM exclusively classified images into these three categories, both YOLOV8 and RF3 extended their capabilities to detect normal colonic tissue, polyps, and resected polyps, with YOLOV8 and RF3 achieving overall accuracies of 0.84 and 0.87, respectively. (4) Conclusions: Machine learning, particularly models like GTM, shows promising results in ensuring comprehensive detection of polyps during colonoscopies.

Strongly Coupled Fluid-Structure Interaction Cardiovascular Analysis with the Effect of Peripheral Network

SEISAN KENKYU, 2011

Additive manufacturing represents a new frontier in the design and production of rotor machines. ... more Additive manufacturing represents a new frontier in the design and production of rotor machines. This technology drives the engineering research framework to new possibilities of design and testing of new prototypes, reducing costs and time. On the other hand, the fast additive manufacturing implies the use of plastic and light materials (as PLA or ABS), often including a certain level of anisotropy due to the layered deposition. These two aspects are critical, because the aero-elastic coupling and flow induced vibrations are not negligible for high aspect ratio rotors. In this work, we investigate the aeroelastic response of a small sample fan blade, printed using PLA material. Scope of the work is to study both the structure and flow field dynamics, where strong coupling is considered on the simulation. We test the blade in two operating points, to see the aero-mechanical dynamics of the system in stall and normal operating condition. The computational fluid-structure interaction (FSI) technique is applied to simulate the coupled dynamics. The FSI solver is developed on the base of the finite element stabilized formulations proposed by Tezduyar et al. We use the ALE formulation of RBVMS-SUPS equations for the aerodynamics, the non-linear elasticity is solved with the Updated Lagrangian formulation of the equations of motion for the elastic solid. The strong coupling is made with a block-iterative algorithm, including the Jacobian based stiffness method for the mesh motion.

Thresholding Segmentation Errors and Uncertainty with Patient-Specific Geometries

Journal of biomedical physics & engineering, Feb 1, 2021

Computer simulations provide virtual hands-on experience when actual hands-on experience is not p... more Computer simulations provide virtual hands-on experience when actual hands-on experience is not possible. To use these simulations in medical science, they need to be able to predict the behavior of actual processes with actual patient-specific geometries. Many uncertainties enter in the process of developing these simulations, starting with creating the geometry. The actual patient-specific geometry is often complex and hard to process. Usually, simplifications to the geometry are introduced in exchange for faster results. However, when simplified, these simulations can no longer be considered patient-specific as they do not represent the actual patient they come from. The ultimate goal is to keep the geometries truly patient-specific without any simplification. However, even without simplifications, the patient-specific geometries are based on medical imaging modalities and consequent use of numerical algorithms to create and process the 3D surface. Multiple users are asked to process medical images of a complex geometry. Their resulting geometries are used to assess how the user's choices determine the resulting dimensions of the 3D model. It is shown that the resulting geometry heavily depends on user's choices.

Use of trefftz functions in modelling of point and line contact

Publisher Summary This chapter contains description of Trefftz functions as a test function for s... more Publisher Summary This chapter contains description of Trefftz functions as a test function for solution of contact problems of bodies with curved surfaces, such as rolling elements, intrusions, and cavities with smooth shape. Trefftz functions are functions that satisfy all governing equations in the whole domain and can be in the form of polynomials, harmonic, or exponential. Since Trefftz functions are regular, much less requirements are put on numerical integration than in the conventional boundary integral method. In this chapter, the use of Trefftz functions, especially Trefftz polynomials, Kelvin/Kupradze's, and Boussinesq functions are presented. The models described in this chapter are an effective way not only for the solution of contact problems (especially in modelling of the multibody contact) but also for some other types of problems such as cracks in bodies.

Modeling Dynamics of the Cardiovascular System Using Fluid-Structure Interaction Methods

Biology

Using fluid-structure interaction algorithms to simulate the human circulatory system is an innov... more Using fluid-structure interaction algorithms to simulate the human circulatory system is an innovative approach that can provide valuable insights into cardiovascular dynamics. Fluid-structure interaction algorithms enable us to couple simulations of blood flow and mechanical responses of the blood vessels while taking into account interactions between fluid dynamics and structural behaviors of vessel walls, heart walls, or valves. In the context of the human circulatory system, these algorithms offer a more comprehensive representation by considering the complex interplay between blood flow and the elasticity of blood vessels. Algorithms that simulate fluid flow dynamics and the resulting forces exerted on vessel walls can capture phenomena such as wall deformation, arterial compliance, and the propagation of pressure waves throughout the cardiovascular system. These models enhance the understanding of vasculature properties in human anatomy. The utilization of fluid-structure inte...

Fluid-structure interaction analysis of amniotic fluid with fetus and placenta inside uterus exposed to military blasts

Injury

Cushioning Effect of Conventional Padded Helmets on Interaction between Cerebrospinal Fluid and Brain after a Low-Speed Head Impact

Applied sciences, Apr 3, 2023

This article is an open access article distributed under the terms and conditions of the Creative... more

Understanding Traumatic Brain Injuries in Military Personnel: Investigating the Dynamic Interplay of the Cerebrospinal Fluid and Brain During Blasts

Cureus, Oct 12, 2023

Background It is estimated that around 450,000 traumatic brain injury cases have occurred in the ... more Background It is estimated that around 450,000 traumatic brain injury cases have occurred in the 21st century with possible under-reporting. Computational simulations are increasingly used to study the pathophysiology of traumatic brain injuries among US military personnel. This approach allows for investigation without ethical concerns surrounding live subject testing. Methodology The pertinent data on head acceleration is applied to a detailed 3D model of a patient-specific head, which encompasses all significant components of the brain and its surrounding fluid. The use of finite element analysis and smoothed-particle hydrodynamics serves to replicate the interaction between these elements during discharge through simulation of their fluid-structure dynamics. Results The stress levels of the brain are assessed at varying time intervals subsequent to the explosion. The regions where there is an intersection between the skull and brain are observed, along with the predominant orientations in which displacement of the brain occurs resulting in a brain injury. Conclusions It has been determined that the cerebrospinal fluid is inadequate in preventing brain damage caused by multiple abrupt directional shifts of the head. Accordingly, additional research must be undertaken to enhance our comprehension of the injury mechanisms linked with consecutive changes in acceleration impacting the head.

Engineering in Medicine: Bridging the Cognitive and Emotional Distance between Medical and Non-Medical Students

International Journal of Education in Mathematics, Science and Technology

In the current study, we focused on measuring the development of important professional attitudes... more In the current study, we focused on measuring the development of important professional attitudes, such as “compassion satisfaction” and “burnout”. Students from four different colleges worked in teams to conceptualize innovative engineering products. During the ideation phase of their project, participants completed a Professional Quality of Life survey to assess metrics related to compassion satisfaction and burnout. On average, the combined compassion satisfaction score was high for both medical students (42/50) and non-medical students (43/50). In terms of burnout, 77% of medical students and 81% of non-medical students reported low burnout; the average burnout score for medical students was 19/50, and for non-medical students 17/50. Only one statement produced a statistically significant difference between groups. For the statement, “I am a caring person”, only 31% of medical students self-described as being a very caring person ‘very often’ as opposed to 62% of non-medical ...

Comparative Analysis of Machine Learning Models for Image Detection of Colonic Polyps vs. Resected Polyps

Journal of Imaging

This article is an open access article distributed under the terms and conditions of the Creative... more

Predictive Modeling in Medicine

Encyclopedia

Predictive modeling is a complex methodology that involves leveraging advanced mathematical and c... more Predictive modeling is a complex methodology that involves leveraging advanced mathematical and computational techniques to forecast future occurrences or outcomes. This tool has numerous applications in medicine, yet its full potential remains untapped within this field. Therefore, it is imperative to delve deeper into the benefits and drawbacks associated with utilizing predictive modeling in medicine for a more comprehensive understanding of how this approach may be effectively leveraged for improved patient care. When implemented successfully, predictive modeling has yielded impressive results across various medical specialities. From predicting disease progression to identifying high-risk patients who require early intervention, there are countless examples of successful implementations of this approach within healthcare settings worldwide. However, despite these successes, significant challenges remain for practitioners when applying predictive models to real-world scenarios. ...

Management of Intracranial Hemorrhage in the Setting of Mechanical Heart Valve Replacement Therapy

Applied Mechanics

The management of an intracranial hemorrhage in patients receiving anticoagulant therapy presents... more The management of an intracranial hemorrhage in patients receiving anticoagulant therapy presents a significant challenge for medical professionals. Anticoagulant treatment is intended to prevent blood clotting, but it can worsen active brain bleeds. Despite this risk, avoiding the prothrombotic state caused by mechanical heart valves remains crucial. Guidelines on managing this issue are currently lacking, prompting a review that delves into embryonic development and anatomical functions of heart valves, valve replacement therapy for diseased valves, and the need for anticoagulants. Ultimately, recent literature and cases inform discussion regarding how best to manage intracranial hemorrhages in patients with mechanical heart valves. The expectation is that this examination will offer valuable perspectives on the handling of intracranial bleeding among individuals with mechanical heart valves and stimulate additional investigations in this intricate domain, particularly through the...

Fluid-Structure Interaction Analyses of Amniotic Fluid with a Comprehensive Fetus Model Exposed to External Loading

8th European Congress on Computational Methods in Applied Sciences and Engineering

Assessment of Abdominal Constrictor’s Forces for Informing Computational Models of Orthostatic Hypotension

Materials

Orthostatic hypotension is defined as a sudden drop in blood pressure upon standing from a sittin... more Orthostatic hypotension is defined as a sudden drop in blood pressure upon standing from a sitting or supine position. The prevalence of this condition increases exponentially with age. Nonpharmacological treatments are always the first step in the management of this condition, such as the use of an abdominal constriction belt to optimize the blood volume in the abdomen. A multitude of clinical trials have shown the efficacy of elastic abdominal compression as well as compression using an inflatable bladder; however, there are currently few accessible consumer products that can provide abdominal compression by using an inflatable bladder that ensures the correct amount of pressure is being exerted on the subject. This study serves to quantitatively analyze forces exerted in inflatable abdominal binders, a novel treatment that fits the criterion for a first-line intervention for orthostatic hypotension. Quantitative values aim to indicate both the anatomic regions of the body subject...

Instituto Superior técnico Modelling of Hydrated Soft Tissues Using Hybrid-Trefftz Finite Elements

A formulação híbrida-Trefftz do método dos elementos finitos é aplicada à análise transiente, geo... more A formulação híbrida-Trefftz do método dos elementos finitos é aplicada à análise transiente, geométrica e fisicamente linear, de meios incompressíveis bifásicos saturados. São desenvolvidos dois modelos, o modelo de tensão e o modelo de deslocamento, os quais são aplicados e testados na análise de problemas bidimensionais e axisimétricos. Realizam-se, para cada um dos casos, dois tipos de análise, no domínio da frequência e no domínio no tempo. O modelo de tensão (deslocamento) decorre da aproximação directa do campo de tensões e de pressão (de deslocamento) em cada fase da mistura no domínio do elemento e dos deslocamentos (das forças e da pressão) nas fases sólida e líquida da fronteira do elemento. As bases de aproximação no domínio usadas na implementação dos modelos são extraídas da solução formal do sistema de equações diferenciais que governa o comportamento da mistura. Os modelos são deduzidos directamente das condições fundamentais de equilíbrio, compatibilidade, elasticid...