Search published articles
Volume 67, Issue 6 (9-2009)
Background: Data mining as a multidisciplinary field is rooted in the fields such as statistics, mathematics, computer science and artificial intelligence and has been gaining momentum in scientific, managerial, and executive applications in health care. Data mining can be defined as the automated extraction of valuable, practical and hidden knowledge and information from large data. Applying data mining in medical records and data is of utmost importance for health care givers and providers and brings vital and valuable outcomes. Data mining can help doctors come up with better recommendations and plans for treatment which actually in many respects have significant impact on patients’ life and satisfaction In this paper we have proposed and utilized data mining methods to extract hidden information in medical records of pelvis stone patients with ureteral stone. We have tried to design a decision support system model to be applicable for selecting type of treatment for these groups of patients.
Methods: We gathered needed information from Shahid Hashemi Nejad hospital. In this research we have used decision tree as a data mining tool, for selecting suitable treatment for patients with ureteral stone. This model can predict probability of success of each treatment.
Results: In this research we extracted effective attributes in selecting type of treatment for patients with ureteral stone.
Conclusions: By using this model we can have eight percent improvement in number of patients who have stone free output after treating. In fact, this model has a better functionality than expert system of hospitals.
Volume 73, Issue 6 (9-2015)
Background: Patient-centered care improves the quality of life and health care, and reduces the costs of care. The advent of new technologies such as health social networks, and personal health records (PHR), have significant impact on the patient-centered care. The aim of this article is to analyze and provide a set of features and requirements needed by the users of health social network serving as a PHR (Personal Health Record) system. The combination of capabilities offered by PHRs and social networks providing better delivery of patient-centered care. Methods: In this paper, after a brief study of capabilities and features of existing health social networks and based on a comparative study, a set of requirements which are necessary to create a comprehensive health social network as a PHR system are proposed. Identification of such a systems stakeholders and users e.g. healthcare professionals, patients, and healthcare organizations is important for categorizing the requirements. Also, classifying relatively vast range of existing systems is needed to have a better analyze and design. Results: The proposed health social network can be used by different user groups in healthcare e.g. healthcare professionals, patients, and healthcare organizations. According to the each user group’s requirements, it provides separate facilities for them. The users of this integrated health social network can optionally share some of their information with other users in their group or with users in the other groups and interact with them. Studies show that the proposal requirements and capabilities for health social network not only cover the capabilities of similar systems but also satisfactory provide the requirements of a PHR system to deliver the patient-centered care. Conclusion: The proposed set of requirements are qualitatively compared with the other similar systems. Using the proposed health social network that provides PHR capabilities for its users will have an irrefutable impact on quality and efficiency of patient-centered care, and play an important role in improving the health of society.
Volume 74, Issue 1 (4-2016)
Background: Data mining is known as a process of discovering and analysing large amounts of data in order to find meaningful rules and trends. In healthcare, data mining offers numerous opportunities to study the unknown patterns in a data set. These patterns can be used to diagnosis, prognosis and treatment of patients by physicians. The main objective of this study was to predict the level of serum ferritin in women with anemia and to specify the basic predictive factors of iron deficiency anemia using data mining techniques.
Methods: In this research 690 patients and 22 variables have been studied in women population with anemia. These data include 11 laboratories and 11 clinical variables of patients related to the patients who have referred to the laboratory of Imam Hossein and Shohada-E- Haft Tir hospitals from April 2013 to April 2014. Decision tree technique has been used to build the model.
Results: The accuracy of the decision tree with all the variables is 75%. Different combinations of variables were examined in order to determine the best model to predict. Regarding the optimum obtained model of the decision tree, the RBC, MCH, MCHC, gastrointestinal cancer and gastrointestinal ulcer were identified as the most important predictive factors. The results indicate if the values of MCV, MCHC and MCH variables are normal and the value of RBC variable is lower than normal limitation, it is diagnosed that the patient is likely 90% iron deficiency anemia.
Conclusion: Regarding the simplicity and the low cost of the complete blood count examination, the model of decision tree was taken into consideration to diagnose iron deficiency anemia in patients. Also the impact of new factors such as gastrointestinal hemorrhoids, gastrointestinal surgeries, different gastrointestinal diseases and gastrointestinal ulcers are considered in this paper while the previous studies have been limited only to assess laboratory variables. The rules of the decision tree model can improve the process of diagnosing and treatment of the patients with iron deficiency anemia and reduce their costs.
Volume 74, Issue 2 (5-2016)
Background: Anemia disease is the most common hematological disorder which most often occurs in women. Knowledge discovery from large volumes of data associated with records of the disease can improve medical services quality by data mining The goal of this study was to determining and evaluating the status of anemia using data mining algorithms.
Methods: In this applied study, laboratory and clinical data of the patients with anemia were studied in the population of women. The data have been gathered during a year in the laboratory of Imam Hossein and Shohada-ye Haft-e Tir Hospitals which contains 690 records and 15 laboratory and clinical features of anemia. To discover hidden relationships and structures using k-medoids algorithm the patients were clustered. The Silhouette index was used to determine clustering quality.
Results: The features of red blood cell (RBC), mean corpuscular hemoglobin (MCH), ferritin, gastrointestinal cancer (GI cancer), gastrointestinal surgery (GI surgery) and gastrointestinal infection (GI infection) by clustering have been determined as the most important patients’ features. These patients according to their features have been seg-mented to three clusters. First, the patients were clustered according to all features. The results showed that clustering with all features is not suitable because of weak structure of clustering. Then, each time the clustering was performed with different number of features. The silhouette index average is 80 percent that shows clustering quality. Therefore clustering is acceptable and has a strong structure.
Conclusion: The results showed that clustering with all features is not suitable because of weak structure. Then, each time the clustering was performed with different number of features. The first cluster contains mild iron deficiency anemia, the second cluster contains severe iron deficiency anemia patients and the third cluster contains patients with other anemia cause.
Volume 75, Issue 1 (4-2017)
|
Background: One of the today most common and incurable diseases that is associated with central neural system is ‘MS’ disease. Multiple sclerosis (MS) is a demyelinating disease in which the insulating covers of nerve cells in the brain and spinal cord are damaged. In this disease become apparent a wide spectrum of symptoms such as lose muscles control and their coordination and vision derangement. The goal of this research is to consider to two problems: 1- Recognition of effective clinical symptoms on MS disease and 2- Considering levels of effectiveness of age, sex and education levels factors on MS disease and association between these factors according to verity of categories of this disease. Methods: Data mining science in medicine is worthy of attention with main application in diagnosis, therapy and prognosis, respectively high volume of collected datum. The data that were used in this article are about patients of Chaharmahal and Bakhtiari Province and collected by cure assistance. In this paper classification and association methods in software engineering field are used. Classification is a general process related to categorization, the process in which ideas and objects are recognized, differentiated, and understood. Association rules are created by analyzing data for frequent if/then patterns and using the criteria support and confidence to identify the most important relationships. Results: In consideration of first problem in this paper, concluded vision-clinical symptoms are the most effective symptoms and in consideration of second problem, concluded that from 584 records, women affected four times more than men. In other word 70% of MS patients with high graduate are in relapsing-remitting category and 62.5% of MS patients are 20-40 years old. Conclusion: Some of symptoms are quite temporary and transitory and are ignored by people. Awareness of clinical-symptoms prevalence manner can be warning for people before starting critical cycle of illness. This would cause early diagnosis, effective therapy and even prevention of disease progress, respectively to MS chronicity. |
Volume 75, Issue 10 (1-2018)
Methods: In this retrospective study, according to the difference of relative importance of demographic characteristics, laboratory data, dialysis adequacy parameters and nutritional status in various patients, the factors affecting the survival of peritoneal dialysis patients have been identified by random forest algorithm. Then, the clinical and laboratory data of patients undergoing continuous ambulatory peritoneal dialysis treatment were evaluated retrospectively from July 1996 to April 2014 in 18 peritoneal dialysis centers, using multi-class one against all support vector machine (OAA-SVM) and multi-space mapped binary tree support vector machine (MBT-SVM) algorithms.
Results: 3097 patients were studied with the mean age of 50.63±15.67 years and average follow-up time of 24.48±19.13 months. The results of the random forest algorithm have identified 35 factors as the most important predictors of peritoneal dialysis patient’s survival. Then, the prediction of peritoneal dialysis patients’ survival status was evaluated using one against all support vector machine and multi-space mapped binary tree support vector machine algorithms in 5 classes of patients including “still on peritoneal dialysis”, “transferred to hemodialysis”, “received a kidney transplant”, “died” and “improved kidney function”. The reliability of survival prediction algorithms were 51.99% and 89.57% respectively.
Conclusion: An accurate prediction model would be a potentially useful way to evaluate patients’ survival at peritoneal dialysis that increased clinical scrutiny and timely intervention could be brought to bear. So, in this research, the multi-space mapped binary tree support vector machine algorithm has a high precision in predicting the survival of continuous ambulatory peritoneal dialysis patients considering multiple evaluation indices and different class distribution functions.
Volume 76, Issue 12 (3-2019)
Methods: This is a descriptive-analytic study that was conducted as a cross-sectional study. The study population included people referring to health centers in Mohammadieh City in Qazvin Province, Iran, from April to June 2015 for screening for diabetes. The 5-step CRISP method was used to implement this study. Data were collected from March 2015 to June 2015. In this study, 1055 persons with complete information were included in the study. Of these, 159 were healthy and 896 were diabetic. A total of 11 characteristics and risk factors were examined, including the age, sex, systolic and diastolic blood pressure, family history of diabetes, BMI, height, weight, waistline, hip circumference and diagnosis. The results obtained by support vector machine (SVM), decision tree (DT) and the k-nearest neighbors algorithm (k-NN) were compared with each other. Data was analyzed using MATLAB® software, version 3.2 (Mathworks Inc., Natick, MA, USA).
Results: Data analysis showed that in all criteria, the best results were obtained by decision tree with accuracy (0.96) and precision (0.89). The k-NN methods were followed by accuracy (0.96) and precision (0.83) and support vector machine with accuracy (0.94) and precision (0.85). Also, in this study, decision tree model obtained the highest degree of class accuracy for both diabetes classes and healthy in the analysis of confusion matrix.
Conclusion: Based on the results, the decision tree represents the best results in the class of test samples which can be recommended as a model for predicting diabetes type 2 using risk factor data.
Volume 77, Issue 5 (8-2019)
Methods: In this cross-sectional study, the data of 7,000 participants in the Diabetes Screening Project in Samen, Mashhad City, Iran, were considered in 2016. There were 540 untreated diabetic patients. The Samen Project was included in the routine examinations of diabetes patients like blood glucose, eyes health, nephropathy, and legs health. So, in order to maintain balance, 600 healthy individuals were selected in a proportional volume sampling in this study. Therefore, the total sample size was 1140 people. In this study, people with diabetes aged over 30 years old were enrolled and participants with the previous history of type 2 diabetes, with normal blood glucose due to drug use or other issues at the time of the study, were excluded.
Results: All three models (Logistic regression, simple Bayesian and support vector machine models) had the same test accuracy (86%), however, in terms of area under the receiver operating characteristic (ROC) curve (AUC), logistic regression and simple Bayesian models had better performance (AUC=90% against AUC=88%). In the simple Bayesian model and logistic regression, body mass index (BMI) and age variables were the most important variables, while BMI and blood pressure variables were the most important factors in the support vector machine model.
Conclusion: According to the results, all three models had the same accuracy. In terms of area under the curve (AUC), logistic and simple Bayes models had better performance than the support vector machine model. Totally all three models had almost the same performance. Based on all three models, BMI was the most important variable.
Volume 79, Issue 3 (6-2021)
|
|
Volume 79, Issue 6 (9-2021)
Background: Millions of deaths occur around the world each year due to lack of access to appropriate treatment for chronic kidney disease patients. Given the importance and mortality rate of this disease, early and low-cost prediction is very important. The researchers intend to identify chronic kidney disease through the optimal combination of techniques used in different stages of data mining.
Methods: This cross-sectional research was conducted from February 1999 to May 2014. The used data set included 4145 samples and 32 attributes, where Each sample corresponded to a patient and each attribute corresponded to the demographic and clinical traits. There were several eligibility criteria for the patients for clinical testing. These criteria for the clinical testing included having 18 years of age and older, living in Isfahan city, willing to participate in the study, lack of fever and cold during laboratory tests, no strenuous exercise 48 hours before laboratory tests, and fasting. Individuals who had an incomplete questionnaire or were unwilling to perform accurate tests were excluded from the study. The target variable is kidney disease, the values of which include sick and healthy. Four data mining techniques have been used in the dataset. These techniques are support vector machine (SVM), random forest (RF), artificial neural network (ANN) and Chi-square automatic interaction detection (CHAID).
Results: Accuracy is the evaluation criteria for comparing available data mining methods. Based on the accuracy criterion, the support vector machine performed better than other techniques (random forest, neural network and CHAID). The best rule is that if the patients consume salt in their diet, their age is between 50 and 69, and they have diabetes. they are 82% more likely to develop chronic kidney disease.
| Conclusion: The derived rules also showed that if we use salt and we have diabetes, we are at the risk of developing chronic kidney disease. Moreover, having diabetes can increase the risk of mortality in chronic kidney patients. Aged people should also be more careful about getting chronic kidney disease. Because, they are more prone to develop chronic kidney disease. |
Volume 79, Issue 7 (10-2021)
Methods: In this study, among the cases of patients who were admitted to ICUs of the Rasoul Akram and Firoozgar hospitals of Tehran City, Iran, from December 2017 to March 2018, the first 24 hours of the ICUs admission data of 874 cases were gathered. A new model based on the standard methodology CRISP was developed. In the modeling section, two well-known data mining techniques called artificial neural network (ANN), K nearest neighbor (KNN) and decision tree (DT) were used. WEKA 3.9.2 open-source software was implemented for data analysis. Finally, according to the accuracy, sensitivity, specificity criteria and AUC-ROC Curve, the superior model was introduced.
Results: Based on the WEKA results, 19 variables had the most impact on the mortality prediction of patients admitted to ICUs including Glasgow Coma Scale (GCS), mechanical ventilation, surgical service at ICUs admission, gender, temperature, serum creatinine, diabetes, Blood urea nitrogen (BUN), age, addiction, International Normalized Ratio (INR), PH, Partial Thromboplastin Time (PTT), albumin, hemoglobin, glucose, pulse rate, hematocrit (HCT), PO2. Based on the created models, some rules have been extracted which can be used as a pattern to predict the probability of mortality. Although the AUC of the three models was acceptable (KNN 81.5%, ANN 77.5% and DT 74.3%), but the accuracy of decision tree J48 (74.2%) was higher.
|
Conclusion: The study indicated that in the KNN model, the rules derived from it can be effective in mortality prediction in patients admitted to ICUs.
|
Volume 79, Issue 9 (12-2021)
Methods: This cross-sectional research has been done from February 2021 to May 2021. The used medical data set includes information of 353 kidney stone patients in Isfahan. In this study, six target attributes of sodium, phosphate, oxalate, citrate, cysteine and uric acid were identified. The techniques for each of the 6 attributes are used separately. The techniques used in this study were three data mining techniques including random forest (RF), artificial neural network (ANN) and support vector machine (SVM).
|
Results: The best performance in terms of accuracy is related to support vector machine techniques in uric acid class, support vector machine in oxalate class and neural network in cysteine class. The worst performance is related to the random forest technique in the citrate class. The safest rules with a 66% confidence level are for the citrate and sodium classes, and the least reliable rule with a 50% confidence level is for the oxalate class.
|
Volume 79, Issue 10 (1-2022)
Methods: In this study, to obtain high accuracy in diagnosing COVID-19 disease, a complete and effective workflow over data mining methods was proposed, which includes these steps: data pre-analyzing, indication selection, model creation, the measure of performance, and display of results. Data and related indications of patients with COVID-19 were collected from Kerman Afzalipour Hospital and Rafsanjan, Ali Ebn Abi Taleb Hospital. Prediction structures were made and tested via different combinations of the disease indications and seven data mining methods. To discover the best key indications, three criteria including accuracy, validation and F-value were applied and to discover the best data mining methods, accuracy and validation criteria were considered. For each data mining method, the criteria were measured independently and all results were reported for analysis. Finally, the best key indications and data mining methods that can diagnose COVID-19 disease with high accuracy were extracted.
|
Results: 9 key indications and 3 data mining methods were obtained. Experimental results show that the discovered key indications and the best-operating data mining method (i.e. SVM) attain an accuracy of 83.19% for the diagnosis of coronavirus disease.
Conclusion: Due to key indications and data mining methods obtained from this study, it is possible to use this method to diagnose coronavirus disease in different people of different clinical indications with high accuracy. |
Volume 79, Issue 11 (2-2022)
Methods: In the present study, the factors of lymphedema in 1117 patients with breast cancer have been collected. The likelihood of developing lymphedema is predicted using ensemble learning via 5 heterogeneous classification algorithms, feature selection and the genetic algorithm (The Two-layer Ensemble Feature Selection method). After collecting the data of patients with breast cancer from 2009 to 2018, and data preprocessing using the optimized ensemble learning algorithm and feature selection, we will examine the likelihood of developing lymphedema for the new patient. Finally, the factors affecting the disease have been extracted. Excluding the time of collecting statistical data, the period of the study was from September 2019 to February 2021. This study is performed at Seyed Khandan Rehabilitation Center, Tehran, Iran.
Results: The results of algorithms showed that the accuracy of the ensemble learning method with selected classification algorithms (SVM with RBF kernel) is 87% and the accuracy of the ensemble learning with feature selection method is 90%. According to the final evaluation of the proposed method, the most effective risk factors for lymphedema have been extracted.
Conclusion: Unfortunately, treatment and diagnosis are not without complications, and one of the most important of these complications in breast cancer is lymphedema in the upper extremities, which can affect the quality of life in patients. It is essential to have a method that can accurately suggest to a specialist whether a new patient will develop lymphedema in the future or how likely it is to develop it, using patient’s own clinical and demographic characteristics.
Volume 79, Issue 12 (3-2022)
Methods: For this study which is a retrospective cohort, data mining methods in the data set including the information related to 3150 patients who underwent anesthesia and surgery from April 2018 to September 2019 in Imam Khomeini Hospital in Kermanshah were used. The data set included patients aged 18 years and older (age range of 18 to 96) who underwent a general anesthesia induction test using Propofol and subsequently endotracheal intubation for non-cardiac surgery. If patients did not have intubation, data were missing, or patients underwent intubation after repeated trials, they got excluded. In total, 2640 patients were included in this analysis. Preoperative patient clinical information was collected from pre-anesthesia evaluation records. Intraoperative data were obtained from computer anesthesia records. This data from the patient monitoring system and the anesthesia machine was automatically stored in the anesthesia files, while drug doses and anesthesia techniques were recorded manually. The data were then pre-processed using SPSS software, version 26 (IBM SPSS, Armonk, NY, USA).
|
Results: In this study, 53 features of patients' records were used (The maximum number of features used in previous studies were 48 features, which compared to them, 5 new features were included in the study) for which a P-value was calculated. Finally, features with a P<0.05 (Indicates the level of significance of the variable) were selected. Then, three data mining algorithms, logistic regression, neural networks and decision tree (the most repetitive data mining algorithms based on previous studies) were used to predict blood pressure. Also, using the criteria of accuracy, precision, sensitivity and F function, the performance of three prediction algorithms in data mining was evaluated.
Conclusion: Six features with P<0.05 were selected that the logistic regression model was more accurate, which was presented as the final model for predicting increased blood pressure fluctuations with path coefficients. |
Volume 81, Issue 4 (7-2023)
Methods: This cross-sectional study was conducted from September 2021 to March 2023 for 30 months at Rafsanjan Ali Ebn Abi Taleb Hospital. Predictive models were developed and tested using different combinations of disease characteristics and seven data mining methods in RapidMiner Studio software. The limitations of the study are as follows: 1) The models were based on 40-year-old and older patients records, which may limit the generalization of results to a wider age group. 2) Despite the high accuracy and comprehensiveness of the method, the models were based only on the information of kidney disease patients at Ali Ibn Abi Talib Rafsanjan Hospital. 3) The climate parameter has not been considered in the data set of the investigation to discover the hidden relationships of this parameter with the kidney disease.
|
Results: The results of the experiments in this study showed that the proposed prediction model using the Bayes method and eight identified key features (age, renal biopsy, uremia, sedimentation, albumin, edema, nocturnal enuresis, and urine-specific gravity), can detect kidney disease in people of different clinical characteristics, with 99.38% accuracy.
Conclusion: Considering that the early diagnosis of kidney disease and the adoption of appropriate treatment methods can prevent the progression of kidney damage, in this study, an attempt has been made to achieve this goal by using new statistical methods and artificial intelligence techniques. Based on the proposed method and the conducted experiments, the most important features and the best data mining method were obtained, and finally, kidney disease prediction was possible with high accuracy. |
| Page 1 from 1 |
Related Websites
Site Keywords
Cardiology, Anatomy, Biology, Epidemiology, Immunology, Neuroscience, Physical therapy, Physiology, Dermatology, Endocrinology and metabolicSite Statistics
- Registered users: 7720 users
- Online users: 0 users
- Guest users: 181 users
- All visits: 15222501 visits
- Visits in 24 Hours: 7231 visits
- Total articles: 11209 articles
- Published articles: 4922 articles
© 2024 , Tehran University of Medical Sciences, CC BY-NC 4.0
Designed & Developed by : Yektaweb