machine learning algorithms

Developing an efficient and reliable method for the identification of depression has high importance. The aim of this paper is to propose an approach for depression diagnosis using an interhemispheric asymmetry matrix and machine learning algorithms.

First, EEG signal was acquired from 24 depressed patients and 24 healthy subjects. The EEG signal was acquired from participants for 5 minutes in eyes-closed (EC) and 5 minutes in eyes-open (EO) condition. After preprocessing data, interhemispheric asymmetry for absolute and relative powers of theta and beta frequency bands, theta-to-alpha power ratio, and IAF features were computed. Then, the proposed asymmetry matrix is used as a feature for statistical and classification analysis. In this paper, classification was performed using a support vector machine (SVM), logistic regression, and multi-layer perceptron (MLP). 

The results demonstrated that central and temporal theta absolute power, central and temporal individual alpha frequency (IAF) asymmetries in EC condition and occipital beta absolute power, temporal theta relative power, temporal theta-to-alpha power ratio, and temporal IAF asymmetries in EO condition have significant differences between depressed and healthy groups. Findings show that beta absolute power asymmetry in the occipital region and EO condition is a good biomarker for depression identification with 77.1% accuracy using Gaussian SVM classifier.

The results of this study show performance of proposed asymmetry matrix features in depression detection. Findings show that beta absolute power asymmetry in the occipital region and EO condition is a good biomarker for depression identification.

Dear Editor Traumatic brain injury (TBI) is a common cause of death and disability, with long-term consequences including cognitive impairment, emotional disturbances and physical disability. Prognosis prediction in TBI patients is critical for guiding treatment decisions and improving patient outcomes. The Glasgow coma scale (GCS) and Glasgow outcome scale (GOS) are useful tools to assess and monitor TBI patients, but other factors such as age, motor component of GCS, and type of injury also significantly influence clinical outcomes. With the advent of artificial intelligence (AI), there is growing interest in using machine learning algorithms to predict prognosis in TBI patients.[1] Here we discuss the harnessing of AI in prognosis prediction for TBI patients, as well as its potential benefits, disadvantages, challenges and predicted future directions. Current state of AI in prognosis prediction for TBI patientsReflecting on large bodies of data (clinical, radiographic and laboratory), AI has shown promise in predicting prognosis in TBI patients. Machine learning algorithms (MLA) can identify patterns in data that may not be immediately apparent to human clinicians, allowing for more accurate predictions of patient outcome. There are several studies that have used MLA to develop diagnostic systems for TBI and predict patient outcomes. These studies have used various clinical inputs such as vital signs, electroencephalography (EEG), hospital volume, Charlson comorbidity index and length of stay to develop machine learning (ML) models.[2] In a study conducted by Mawdsley et al, the effectiveness of ML models in predicting psychosocial aspects of TBI cases was systematically reviewed. The study included nine studies with eleven types of ML used to predict various outcomes of traumatic brain injury, concussion and psychosocial outcomes. The results indicated that while these models were able to develop predictive models successfully, there is insufficient evidence to consider ML algorithms as a dependable tool for clinical decision-making.[3] Alanazi et al., conducted a review in 2017 to assess the effectiveness of ML models in predicting patient outcomes for various disorders. The study revealed that AI has the potential to create several promising models that can predict outcomes using clinical, demographic, and imaging data. However, there are still limitations in applying these models in clinical settings. Therefore, further research is necessary to improve the reliability of these models in the future.[4] Potential benefits of AI in prognosis prediction for TBI patientsThe use of AI in prognosis prediction for TBI patients has several potential benefits. First and foremost, it can provide clinicians with more accurate and reliable predictions of patient outcomes, allowing for more informed treatment decisions. This can lead to improved patient outcomes and reduced healthcare costs. AI can help identify patients who are at high risk for poor outcomes, allowing for early interventions that may improve their prognosis, or early decision making around withdrawal of treatment. For example, if an algorithm predicts a patient is at high risk of adverse effects from an intracranial hemorrhage, clinicians can perform early surgical interventions or administer medications early to prevent further bleeding. In addition, AI can help identify patients who are likely to have favorable outcomes, or who do not require intervention, which allows for more efficient distribution of healthcare resources. For example, if an algorithm predicts a patient is likely to have a good outcome, they may not require intensive care unit (ICU) admission, or may not require prolonged rehabilitation services. Finally, AI can also help reduce the workload of medical professionals by performing repetitive tasks such as data entry and analysis. This can free up time for physicians and nurses to focus on providing high-quality care to patients.[1,5] Potential disadvantages of AI in prognosis prediction for TBI patientsThere are several potential disadvantages of using AI in predicting the clinical outcomes of patients with TBI.  AI algorithms rely on large amounts of data to make accurate predictions. However, there may not be enough high-quality data available to train AI models effectively. Data entry relies on human factors and may be poorly or incorrectly entered, making errors possible. Also, AI algorithms may be biased if the data used to train them is not representative of the population being studied. For example, if the data used to train an AI model on TBI outcomes comes from a specific hospital or region, the model may not generalize well to other populations. AI models can be complex and difficult to interpret. Clinicians may struggle to understand how the model arrived at a particular prediction, which could lead to skepticism or mistrust of the technology. Finally, there are ethical considerations around the use of AI in healthcare. If an AI model predicts that a patient has a poor prognosis, this could lead to decisions about end-of-life care, which may go against the patient's wishes or values. Clinicians must carefully consider the potential risks and benefits of using AI in clinical decision-making.[1,5] Future DirectionsThe future of prognosis prediction in TBI patients using AI is promising. However, there are several challenges that must be addressed. One challenge is the lack of standardized data collection and reporting. To develop accurate algorithms, large amounts of high-quality data are required. However, there is currently no standardized way to collect and report TBI data, making it difficult to develop accurate, widely transferable algorithms. Another challenge is the need for validation studies. While initial research has shown promising results, further validation studies are needed to confirm the accuracy and reliability of AI algorithms in predicting TBI prognosis,[3,4] as well as ensuring safety in clinical practice.[1,6]With the increasing availability of electronic health records and the development of advanced AI algorithms, there is potential for more accurate and reliable predictions of TBI outcomes. AI can help identify patterns and correlations in patient data that may not be easily recognizable by human clinicians, leading to more personalized and effective treatment plans, however, there is the challenge of integrating AI into clinical practice. Clinicians must be trained on how to interpret and use AI predictions in their decision-making processes. This will require collaboration between clinicians and AI experts to develop user-friendly interfaces that can be integrated into clinical workflows. ConclusionThe reliability of prognosis prediction in TBI patients using AI depends on the quality and completeness of the data, the type of algorithm used and the level of validation testing performed. Conclusions surrounding the selection of optimal clinical or para-clinical features and the most precise machine learning model for predicting outcomes in TBI patients is still inconsistent. However, early studies have shown that AI algorithms can accurately predict mortality and functional outcomes in TBI patients by analyzing large amounts of patient data, including clinical, radiographic, and laboratory data. Therefore, further validation studies are needed to confirm the accuracy and reliability of AI algorithms in predicting TBI prognosis before they can be integrated into clinical practice. With continued research and collaboration between clinicians and AI experts, the use of AI in prognosis prediction for TBI patients has the potential to revolutionize TBI care and improve patient outcomes.

The purpose of this study is to use linear and non-linear features extracted from Electroencephalography (EEG) signal to predict the Mini-Mental State Examination (MMSE) test score by machine learning algorithms.

First, the MMSE test was taken from 20 subjects that were referred with the initial diagnosis of dementia. Then, the brain activity of subjects was recorded via EEG signal. After preprocessing this signal, various linear and non-linear features are extracted from it that are used as input to machine learning algorithms to predict MMSE test scores in three levels.

Based on the experiments, the best classification result is related to the Long Short-Term Memory (LSTM) network with 68% accuracy.

Findings show that by using machine learning algorithms and features extracted from EEG signal the MMSE scores are predicted in three levels. Although deep neural networks require a lot of data for training, the LSTM network has been able to achieve the best performance. By increasing the number of subjects, it is expected that the classification results will also increase.

The health industry is a competitive and lucrative industry that has attracted many investors. Therefore, hospitals must create competitive advantages to stay in the competitive market. Patient satisfaction with the services provided in hospitals is one of the most basic competitive advantages of this industry. Therefore, identifying and analyzing the factors affecting the increase of patient satisfaction is an undeniable necessity that has been addressed in this study.

Because patient satisfaction characteristics used in hospitals may have a hidden relationship with each other, data mining approaches and tools to analyze patient satisfaction according to the questionnaire used We used the hospital. After preparing the data, the characteristics mentioned in the questionnaire for patients, classification models were applied to the collected and cleared data, and with the feature selection methods, effective characteristics Patients were identified and analyzed for satisfaction or dissatisfaction.

Based on the findings of the present study, it can be concluded that the factors of patient mentality of the physician's expertise and skill, appropriate and patient behavior of the physician and food quality (hoteling) respectively have a higher chance of increasing patient satisfaction with Establish services provided in the hospital.

Comparing the approach used in this study with other studies showed that due to the hidden effects of variables on each other and the relatively large number of variables studied, one of the best options for analyzing patient satisfaction questionnaire data, Use of data mining tools and approaches

Nucleic acid-binding proteins play major roles in different biological processes, such as transcription, splicing and translation. Therefore, the nucleic acid-binding function prediction of proteins is a step toward full functional annotation of proteins. The aim of our research was the improvement of nucleic-acid binding function prediction. In the current study, nine machine-learning algorithms were used to predict RNA- and DNA-binding proteins and also to discriminate between RNA-binding proteins and DNA-binding proteins. The electrostatic features were utilized for prediction of each function in corresponding adapted protein datasets. The leave-one-out cross-validation process was used to measure the performance of employed classifiers. Radial basis function classifier gave the best results in predicting RNA- and DNA-binding proteins in comparison with other classifiers applied. In discriminating between RNA- and DNA-binding proteins, multilayer perceptron classifier was the best one. Our findings show that the prediction of nucleic acid-binding function based on these simple electrostatic features can be improved by applied classifiers. Moreover, a reasonable progress to distinguish between RNA- and DNA-binding proteins has been achieved.