Maciej Rosoł

Mr. Maciej Rosoł

orcid

Research Focus & Metrics

Academic Career Length

6Years (2021-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
22 November 2023 in Scientific Reports
Verified
Journal Article
Open access
PDF file
MRMaciej Rosoł
Maciej Rosoł
JGJakub S. Gąsior
Jakub S. Gąsior
JŁJonasz Łaba
Jonasz Łaba
KKKacper Korzeniewski
Kacper Korzeniewski
MMMarcel Młyńczak
Marcel Młyńczak
The study aimed to evaluate the performance of two Large Language Models (LLMs): ChatGPT (based on GPT-3.5) and GPT-4 with two temperature parameter values, on the Polish Medical Final Examination (MFE). The models were tested on three editions of the MFE from: Spring 2022, Autumn 2022, and Spring 2023 in two language versions—English and Polish. The accuracies of both models were compared and the relationships between the correctness of answers with the answer’s metrics were investigated. The study demonstrated that GPT-4 outperformed GPT-3.5 in all three examinations regardless of the language used. GPT-4 achieved mean accuracies of 79.7% for both Polish and English versions, passing all MFE versions. GPT-3.5 had mean accuracies of 54.8% for Polish and 60.3% for English, passing none and 2 of 3 Polish versions for temperature parameter equal to 0 and 1 respectively while passing all English versions regardless of the temperature parameter value. GPT-4 score was mostly lower than the average score of a medical student. There was a statistically significant correlation between the correctness of the answers and the index of difficulty for both models. The overall accuracy of both models was still suboptimal and worse than the average for medical students. This emphasizes the need for further improvements in LLMs before they can be reliably deployed in medical settings. These findings suggest an increasing potential for the usage of LLMs in terms of medical education.

29 January 2022 in Computer Methods and Programs in Biomedicine
Verified
Journal Article
Open access
MRMaciej Rosoł
Maciej Rosoł
MMMarcel Młyńczak
Marcel Młyńczak
GCGerard Cybulski
Gerard Cybulski
Causality defined by Granger in 1969 is a widely used concept, particularly in neuroscience and economics. As there is an increasing interest in nonlinear causality research, a Python package with a neural-network-based causality analysis approach was created. It allows performing causality tests using neural networks based on Long Short-Term Memory (LSTM), Gated Recurrent Unit (GRU), or Multilayer Perceptron (MLP). The aim of this paper is to present the nonlinear method for causality analysis and the created Python package. The created functions with the autoregressive (AR) and Generalized Radial Basis Functions (GRBF) neural network models were tested on simulated signals in two cases: with nonlinear dependency and with absence of causality from Y to X signal. The train-test split (70/30) was used. Errors obtained on the test set were compared using the Wilcoxon signed-rank test to determine the presence of the causality. For the chosen model, the proposed method of study the change of causality over time was presented. In the case when X was a polynomial of Y, nonlinear methods were able to detect the causality, while the AR model did not manage to indicate it. The best results (in terms of the prediction accuracy) were obtained for the MLP for the lag of 150 (MSE equal to 0.011, compared to 0.041 and 0.036 for AR and GRBF, respectively). When there was no causality between the signals, none of the proposed and AR models did indicate false causality, while it was detected by GRBF models in one case. Only the proposed models gave the expected results in each of the tested scenarios. The proposed method appeared to be superior to the compared methods. They were able to detect non-linear causality, make accurate forecasting and not indicate false causality. The created package enables easy usage of neural networks to study the causal relationship between signals. The neural-networks-based approach is a suitable method that allows the detection of a nonlinear causal relationship, which cannot be detected by the classical Granger method. Unlike other similar tools, the package allows for the study of changes in causality over time.

5 June 2023 in medRxiv
Verified
Preprint
Open access
PDF file
MRMaciej Rosoł
Maciej Rosoł
Jakub S. GąsiorJakub S. Gąsior
Jakub S. Gąsior
JŁJonasz Łaba
Jonasz Łaba
KKKacper Korzeniewski
Kacper Korzeniewski
MMMarcel Młyńczak
Marcel Młyńczak
Introduction: The rapid progress in artificial intelligence, machine learning, and natural language processing has led to the emergence of increasingly sophisticated large language models (LLMs) enabling their use in various applications, including medicine and healthcare. Objectives: The study aimed to evaluate the performance of two LLMs: ChatGPT (based on GPT-3.5) and GPT-4, on the Medical Final Examination (MFE). Methods: The models were tested on three editions of the MFE from: Spring 2022, Autumn 2022, and Spring 2023 in two language versions – English and Polish. The accuracies of both models were compared and the relationships between the correctness of answers with the index of difficulty and discrimination power index were investigated. Results: The study demonstrated that GPT-4 outperformed GPT-3.5 in all three examinations regardless of the language used. GPT-4 achieved mean accuracies of 80.7% for Polish and 79.6% for English, passing all MFE versions. GPT-3.5 had mean accuracies of 56.6% for Polish and 58.3% for English, passing 2 of 3 Polish versions and all 3 English versions of the test. GPT-4 score was lower than the average score of a medical student. There was a significant positive and negative correlation between the correctness of the answers and the index of difficulty and discrimination power index, respectively, for both models in all three exams. Conclusions: These findings contribute to the growing body of literature on the utility of LLMs in medicine. They also suggest an increasing potential for the usage of LLMs in terms of medical education and decision-making support. What’s new?: Recent advancements in artificial intelligence and natural language processing have resulted in the development of sophisticated large language models (LLMs). This study focused on the evaluation of the performance of two LLMs, ChatGPT (based on GPT-3.5) and GPT-4, on the Medical Final Examination across English and Polish versions from three editions. This study, to the best of our knowledge, presents the first validation of those models on the European-based medical final examinations. The GPT-4 outperformed GPT-3.5 in all exams, achieving mean accuracy of 80.7% (Polish) and 79.6% (English), while GPT-3.5 attained 56.6% (Polish) and 58.3% (English) respectively. However, GPT-4’s scores fell short of typical medical student performance. These findings contribute to understanding LLM’s utility in medicine and hint at their potential in medical education and decision-making support.

11 January 2021 in Applied Sciences
Verified
Journal Article
Open access
PDF file
MMMarcel Młyńczak
Marcel Młyńczak
MRMaciej Rosoł
Maciej Rosoł
ASAntonino Spinelli
Antonino Spinelli
ADAdam Dziki
Adam Dziki
EWEdyta Wlaźlak
Edyta Wlaźlak
KBKatarzyna Borycka
Katarzyna Borycka
Anal sphincter injuries occurring during natural deliveries are often a reason for severe complications, including fecal incontinence. Currently, approximately 80% of these injuries remain unrecognized. Therefore, it is crucial to focus on finding a way to diagnose such injuries as early as possible to apply the tailored treatment. This study aimed to assess the accuracy of impedance spectroscopy in the diagnostics of obstetric anal sphincter injuries (OASIs) using a specially designed rectal probe called the ONIRY Probe. The protocol of the clinical trials is described at NCT03769792. Twenty women after natural delivery were enrolled in the study and divided into two groups referring to the stage of a perineal tear (Group A: 1- or 2-degrees, and Group B: 3- or 4-degrees of a perineal tear; without or with sphincter injury, respectively). The study design included three visits during which a number of diagnostic tests were performed, including impedance spectroscopy, and 3-dimensional endoanal ultrasound, anorectal manometry, and physical examination as a reference. Statistical analysis comprised raw data analysis, as well as post-processing with the Synthetic Minority Oversampling Technique (SMOTE) method, as the output reference grade was highly imbalanced. A variety of machine learning techniques were applied for the OASI classification (≤2 vs. >2), and Wexner scale (=0 vs. >0). The best efficacies were obtained using Random Forest and k-Nearest Neighbors methods. Best accuracies were 93.3% and 99.6%, for raw and re-analyzed data, respectively, for ultrasound assessed by the OASI classification; and 79.8% and 97.0%, respectively, for clinical evaluation using the Wexner scale. Impedance spectroscopy performed using the ONIRY Probe appears to be a promising diagnostic technique for anal sphincter injury detection but requires further investigation (the next phase of the pilot clinical trial is described at NCT04181840).

10 January 2024 in PLOS ONE
Verified
Journal Article
Open access
PDF file
MRMaciej Rosoł
Maciej Rosoł
MPMonika Petelczyc
Monika Petelczyc
Jakub S. GąsiorJakub S. Gąsior
Jakub S. Gąsior
MMMarcel Młyńczak
Marcel Młyńczak
This study investigates the quality of peak oxygen consumption (VO2peak) prediction based on cardiac and respiratory parameters calculated from warmup and submaximal stages of treadmill cardiopulmonary exercise test (CPET) using machine learning (ML) techniques and assesses the importance of respiratory parameters for the prediction outcome. The database consists of the following parameters: heart rate (HR), respiratory rate (RespRate), pulmonary ventilation (VE), oxygen consumption (VO2) and carbon dioxide production (VCO2) obtained from 369 treadmill CPETs. Combinations of features calculated based on the HR, VE and RespRate time-series from different stages of CPET were used to create 11 datasets for VO2peakprediction. Thirteen ML algorithms were employed, and model performances were evaluated using cross-validation with mean absolute percentage error (MAPE), R2score, mean absolute error (MAE), and root mean squared error (RMSE) calculated after each iteration of the validation. The results demonstrated that incorporating respiratory-based features improves the prediction of VO2peak. The best results in terms of R2score (0.47) and RMSE (5.78) were obtained for the dataset which included both cardiac- and respiratory-based features from CPET up to 85% of age-predicted HRmax, while the best results in terms of MAPE (10.5%) and MAE (4.63) were obtained for the dataset containing cardiorespiratory features from the last 30 seconds of warmup. The study showed the potential of using ML models based on cardiorespiratory features from submaximal tests for prediction of VO2peakand highlights the importance of the monitoring of respiratory signals, enabling to include respiratory parameters into the analysis. Presented approach offers a feasible alternative to direct VO2peakmeasurement, especially when specialized equipment is limited or unavailable.

8 August 2024 in Journal of Clinical Medicine
Verified
Journal Article
Open access
PDF file
Jakub S. GąsiorJakub S. Gąsior
Jakub S. Gąsior
MMMarcel Młyńczak
Marcel Młyńczak
MRMaciej Rosoł
Maciej Rosoł
PWPiotr Wieniawski
Piotr Wieniawski
RPRadosław Pietrzak
Radosław Pietrzak
Bożena WernerBożena Werner
Bożena Werner
Background: Wearable technologies have been developed to measure physiological parameters conveniently. To consider the new measurement device valid, the crucial point is to assess its reliability with the gold standard. The study aimed to assess the validity of the Pneumonitor (PM, fs = 250 Hz) for acquisition of 5 min RR intervals (RRi) for analysis of heart rate asymmetry (HRA) in relation to the electrocardiography (ECG, fs = 1000 Hz) in a group of 19 pediatric cardiac patients. Association between HRA and respiratory rate (RespRate) was verified. Methods: The validation comprised Bland–Altman analysis, intraclass correlation coefficient, and Student’s t-test. Results: Sufficient agreement between 10 from 16 HRA parameters was observed. Different HRA parameters values calculated based on RRi from both devices were related to different results of correlation analysis between two parameters and RespRate. Conclusions: The PM might be considered valid for recording RRi, which are then processed to calculate selected HRA parameters in a group of pediatric cardiac patients in rest condition. However, RRi recorded using devices with fs < 250 Hz may be not adequate for reliable HRA analysis.