MRI-based brain tumor detection using convolutional deep learning methods and chosen machine learning techniques.

BACKGROUND: Detecting brain tumors in their early stages is crucial. Brain tumors are classified by biopsy, which can only be performed through definitive brain surgery. Computational intelligence-oriented techniques can help physicians identify and classify brain tumors. Herein, we proposed two deep learning methods and several machine learning approaches for diagnosing three types of tumor, i.e., glioma, meningioma, and pituitary gland tumors, as well as healthy brains without tumors, using magnetic resonance brain images to enable physicians to detect with high accuracy tumors in early stages. MATERIALS AND METHODS: A dataset containing 3264 Magnetic Resonance Imaging (MRI) brain images comprising images of glioma, meningioma, pituitary gland tumors, and healthy brains were used in this study. First, preprocessing and augmentation algorithms were applied to MRI brain images. Next, we developed a new 2D Convolutional Neural Network (CNN) and a convolutional auto-encoder network, both of which were already trained by our assigned hyperparameters. Then 2D CNN includes several convolution layers; all layers in this hierarchical network have a 2*2 kernel function. This network consists of eight convolutional and four pooling layers, and after all convolution layers, batch-normalization layers were applied. The modified auto-encoder network includes a convolutional auto-encoder network and a convolutional network for classification that uses the last output encoder layer of the first part. Furthermore, six machine-learning techniques that were applied to classify brain tumors were also compared in this study. RESULTS: The training accuracy of the proposed 2D CNN and that of the proposed auto-encoder network were found to be 96.47% and 95.63%, respectively. The average recall values for the 2D CNN and auto-encoder networks were 95% and 94%, respectively. The areas under the ROC curve for both networks were 0.99 or 1. Among applied machine learning methods, Multilayer Perceptron (MLP) (28%) and K-Nearest Neighbors (KNN) (86%) achieved the lowest and highest accuracy rates, respectively. Statistical tests showed a significant difference between the means of the two methods developed in this study and several machine learning methods (p-value < 0.05). CONCLUSION: The present study shows that the proposed 2D CNN has optimal accuracy in classifying brain tumors. Comparing the performance of various CNNs and machine learning methods in diagnosing three types of brain tumors revealed that the 2D CNN achieved exemplary performance and optimal execution time without latency. This proposed network is less complex than the auto-encoder network and can be employed by radiologists and physicians in clinical systems for brain tumor detection.

Prediction of COVID-19 Patients' Survival by Deep Learning Approaches.

Background: Despite many studies done to predict severe coronavirus 2019 (COVID-19) patients, there is no applicable clinical prediction model to predict and distinguish severe patients early. Based on laboratory and demographic data, we have developed and validated a deep learning model to predict survival and assist in the triage of COVID-19 patients in the early stages. Methods: This retrospective study developed a survival prediction model based on the deep learning method using demographic and laboratory data. The database consisted of data from 487 patients with COVID-19 diagnosed by the reverse transcription-polymerase chain reaction test and admitted to Imam Khomeini hospital affiliated to Tehran University of Medical Sciences from February 21, 2020, to June 24, 2020. Results: The developed model achieved an area under the curve (AUC) of 0.96 for survival prediction. The results demonstrated the developed model provided high precision (0.95, 0.93), recall (0.90,0.97), and F1-score (0.93,0.95) for low- and high-risk groups. Conclusion: The developed model is a deep learning-based, data-driven prediction tool that can predict the survival of COVID-19 patients with an AUC of 0.96. This model helps classify admitted patients into low-risk and high-risk groups and helps triage patients in the early stages.

Digital Health Solutions to Control the COVID-19 Pandemic in Countries With High Disease Prevalence: Literature Review.

BACKGROUND: COVID-19, the disease caused by the novel coronavirus SARS-CoV-2, has become a global pandemic, affecting most countries worldwide. Digital health information technologies can be applied in three aspects, namely digital patients, digital devices, and digital clinics, and could be useful in fighting the COVID-19 pandemic. OBJECTIVE: Recent reviews have examined the role of digital health in controlling COVID-19 to identify the potential of digital health interventions to fight the disease. However, this study aims to review and analyze the digital technology that is being applied to control the COVID-19 pandemic in the 10 countries with the highest prevalence of the disease. METHODS: For this review, the Google Scholar, PubMed, Web of Science, and Scopus databases were searched in August 2020 to retrieve publications from December 2019 to March 15, 2020. Furthermore, the Google search engine was used to identify additional applications of digital health for COVID-19 pandemic control. RESULTS: We included 32 papers in this review that reported 37 digital health applications for COVID-19 control. The most common digital health projects to address COVID-19 were telemedicine visits (11/37, 30%). Digital learning packages for informing people about the disease, geographic information systems and quick response code applications for real-time case tracking, and cloud- or mobile-based systems for self-care and patient tracking were in the second rank of digital tool applications (all 7/37, 19%). The projects were deployed in various European countries and in the United States, Australia, and China. CONCLUSIONS: Considering the potential of available information technologies worldwide in the 21st century, particularly in developed countries, it appears that more digital health products with a higher level of intelligence capability remain to be applied for the management of pandemics and health-related crises.

Mobile-Based and Cloud-Based System for Self-management of People With Type 2 Diabetes: Development and Usability Evaluation.

BACKGROUND: As the use of smartphones and mobile apps is increasing, mobile health (mHealth) can be used as a cost-effective option to provide behavioral interventions aimed at educating and promoting self-management for chronic diseases such as diabetes. Although many mobile software apps have been developed for this purpose, they usually lack a theoretical foundation and do not follow the guidelines suggested for evidence-based practice. Therefore, this study aimed to develop a theory-based self-management app for people with type 2 diabetes and provide an app based on a needs assessment analysis. OBJECTIVE: This paper describes the development and usability evaluation of a cloud-based and mobile-based diabetes self-management app designed to help people with diabetes change their health behavior and also enable remote monitoring by health care providers. METHODS: The development of this mHealth solution comprises 3 phases. Phase I: feature extraction of the Android apps that had a user rating of 4 stars or more and review of papers related to mHealth for diabetes self-management were performed followed by seeking expert opinions about the extracted features to determine the essential features of the app. Phase II: design and implementation included selecting which behavioral change and structural theories were to be applied the app and design of the website. Phase III: evaluation of the usability and user experience of the mobile app by people with diabetes and the portal by health care providers using the User Experience Questionnaire. RESULTS: The developed mobile app includes modules that support several features. A person's data were entered or collected and viewed in the form of graphs and tables. The theoretical foundation of behavioral intervention is the transtheoretical model. Users were able to receive customized messages based on the behavioral change preparation stage using the Kreuter algorithm. The clinician's portal was used by health care providers to monitor the patients. The results of the usability evaluation revealed overall user satisfaction with the app. CONCLUSIONS: Mobile- and cloud-based systems may be an effective tool for facilitating the modification of self-management of chronic care. The results of this study showed that the usability of mobile- and cloud-based systems can be satisfactory and promising. Given that the study used a behavioral model, assessment of the effectiveness of behavior change over time requires further research with long-term follow-up.

Effect of Computerized Physician Order Entry and Clinical Decision Support System on Adverse Drug Events Prevention in the Emergency Department: A Systematic Review.

Objective: An adverse drug event (ADE) is an injury resulting from a medical intervention related to a drug. The emergency department (ED) is a ward vulnerable to more ADEs because of overcrowding. Information technologies such as computerized physician order entry (CPOE) and clinical decision support system (CDSS) may decrease the occurrence of ADEs. This study aims to review research that reported the evaluation of the effectiveness of CPOE and CDSS on lowering the occurrence of ADEs in the ED. Data Sources: PubMed, EMBASE, and Web of Science databases were used to find studies published from 2003 to 2018. The search was conducted in November 2018. Study Selection and Data Extraction: The search resulted in 1700 retrieved articles. After applying inclusion and exclusion criteria, 11 articles were included. Data on the date, country, type of system, medication process stages, study design, participants, sample size, and outcomes were extracted. Data Synthesis: Results showed that CPOE and CDSS may prevent ADEs in the ED through significantly decreasing the rate of errors, ADEs, excessive dose, and inappropriate prescribing (in 54.5% of articles); furthermore, CPOE and CDSS may significantly increase the rate of appropriate prescribing and dosing in compliance with established guidelines (45.5% of articles). Conclusion: This study revealed that the use of CPOE and CDSS can lower the occurrence of ADEs in the ED; however, further randomized controlled trials are needed to address the effect of a CDSS, with basic or advanced features, on the occurrence of ADEs in the ED.

Supporting colorectal cancer survivors using eHealth: a systematic review and framework suggestion.

PURPOSE: eHealth could potentially support colorectal cancer survivors; however, little is known regarding the overall recent eHealth systems for colorectal cancer survivors. The present study was conducted to address which types of eHealth supports have been provided to colorectal cancer survivors in the past two decades. METHODS: An electronic search was conducted in four databases including Scopus, PubMed, Embase, and Web of Science. The search query was based on two concepts: the first concept represented colorectal cancer and the second one comprised of information technology tools. The search was limited to 20 years (from 19 January 1999 to 19 January 2019). Obtained results were tabulated and represented as a framework. RESULTS: Fifteen papers were included in this systematic review. Information including intervention type, eHealth tools, main features of the system, and outcomes were extracted from selected papers. Obtained results were characterized using a four-layer framework. This framework included layers of hardware, software, service (educating the patient, medication intake, physical activity, health status monitoring, hospital visit reminder, and discussion group), and outcome. Outcome layer was composed of the following domains: quality of life, psychological and cognitive, physical activity, physical functioning, symptoms, engagement, and the outcome of the process and IT tools. CONCLUSION: eHealth could provide useful services for supporting colorectal cancer survivors. Represented framework might be used for a better understanding of current technology and services provided to support these survivors. Also, this framework may be used as a basis for designing eHealth applications for colorectal cancer survivors after further validations.

Effectiveness of Artificial Intelligence for Personalized Medicine in Neoplasms: A Systematic Review.

Purpose: Artificial intelligence (AI) techniques are used in precision medicine to explore novel genotypes and phenotypes data. The main aims of precision medicine include early diagnosis, screening, and personalized treatment regime for a patient based on genetic-oriented features and characteristics. The main objective of this study was to review AI techniques and their effectiveness in neoplasm precision medicine. Materials and Methods: A comprehensive search was performed in Medline (through PubMed), Scopus, ISI Web of Science, IEEE Xplore, Embase, and Cochrane databases from inception to December 29, 2021, in order to identify the studies that used AI methods for cancer precision medicine and evaluate outcomes of the models. Results: Sixty-three studies were included in this systematic review. The main AI approaches in 17 papers (26.9%) were linear and nonlinear categories (random forest or decision trees), and in 21 citations, rule-based systems and deep learning models were used. Notably, 62% of the articles were done in the United States and China. R package was the most frequent software, and breast and lung cancer were the most selected neoplasms in the papers. Out of 63 papers, in 34 articles, genomic data like gene expression, somatic mutation data, phenotype data, and proteomics with drug-response which is functional data was used as input in AI methods; in 16 papers' (25.3%) drug response, functional data was utilized in personalization of treatment. The maximum values of the assessment indicators such as accuracy, sensitivity, specificity, precision, recall, and area under the curve (AUC) in included studies were 0.99, 1.00, 0.96, 0.98, 0.99, and 0.9929, respectively. Conclusion: The findings showed that in many cases, the use of artificial intelligence methods had effective application in personalized medicine.

A Smartphone Remote Monitoring App to Follow Up Colorectal Cancer Survivors: Requirement Analysis.

BACKGROUND: Colorectal cancer survivors face multiple challenges after discharge. eHealth may potentially support them by providing tools such as smartphone apps. They have lots of capabilities to exchange information and could be used for remote monitoring of these patients. OBJECTIVE: In this study, we addressed the required features for apps designed to follow up colorectal cancer patients based on survivors' and clinical experts' views. METHODS: A mixed methods study was conducted. Features of related apps were extracted through the literature; the features were categorized, and then, they were modified. A questionnaire was designed containing the features listed and prioritized based on the MoSCoW (Must have, Should have, Could have, Won't have) technique and an open question for each category. The link to the questionnaire was shared among clinical experts in Iran. The answers were analyzed using the content validity ratio (CVR), and based on the value of this measure, the minimum feature set of a monitoring app to follow up patients with colorectal cancer was addressed. In addition, a telephone interview with colorectal cancer survivors was conducted to collect their viewpoints regarding a remote monitoring system for colorectal cancer cases. RESULTS: The questionnaire contained 10 sections evaluating 9 categories of features. The questionnaire was completed by 18 experts. The minimum set of features in the app was identified as patient information registration, sign and symptom monitoring, education, reminders, and patient evaluation (0.42 < CVR < 0.85). Features including physical activity, personalized advice, and social network did not achieve the minimum score (-0.11 < CVR < 0.39). We interviewed 9 colorectal cancer survivors. Information registration, sign and symptom monitoring, education, and personalized advice were the features with high priority from the survivors' perspectives. Scheduling, shopping, and financial support features were emphasized by survivors in the interview. CONCLUSIONS: The requirement set could be used to design an app for the targeted population or patients affected by other cancers. As the views from both survivors and clinical experts were considered in this study, the remote system may more adequately fulfill the need for follow-up of survivors. This eases the patients' and health care providers' communication and interaction.

A systematic review of decision aids for mammography screening: Focus on outcomes and characteristics.

PURPOSE: Decision Aid systems (DAs) provide information on the pros and cons of mammography. This study aimed to review the research on mammography DAs, synthesize the findings related to their outcomes and characteristics, and address the existed research gap. METHODS: Relevant studies were identified through a comprehensive search on some e-databases, including PubMed, EMBASE, Scopus, and Web of Science in August 2020; by searching the keywords of "Breast cancer", "Screening", and "Decision aid systems" as well as their synonyms in the titles and abstracts of the papers with no time limits. Among the selected English journal papers with the interventional study design, those measuring outcome values of using mammography DAs were recognized as eligible for being included in this review. RESULTS: The systematic search results in 16 DAs regarding mammography that were designed and then evaluated from 18 selected studies. The results showed that DAs provide improvements in knowledge and informed choice, the decreased decisional conflicts and decisional confidence, almost without changing any attitude towards mammography, mammography participation rates, psychological issues, anticipated regret, and perceived risk of breast cancer. The DAs' effects on women's inclination to screening were divergent. In other words, the DAs affect individuals' inclination in rare cases; however, on occasion, they could affect women's decision to undergo screening. CONCLUSION: DAs could correct the bias attached to the existing knowledge on mammography and breast cancer in women so that they are more likely to make a precise decision. Additionally, it might be of central importance in shared decision-making and assisting health providers, in order to promote the quality of care. Accordingly, performing more studies is needed to develop more professional DAs in various countries with different facilities, cultures, and languages.

Predicting COVID-19 Incidence Through Analysis of Google Trends Data in Iran: Data Mining and Deep Learning Pilot Study.

BACKGROUND: The recent global outbreak of coronavirus disease (COVID-19) is affecting many countries worldwide. Iran is one of the top 10 most affected countries. Search engines provide useful data from populations, and these data might be useful to analyze epidemics. Utilizing data mining methods on electronic resources' data might provide a better insight into the COVID-19 outbreak to manage the health crisis in each country and worldwide. OBJECTIVE: This study aimed to predict the incidence of COVID-19 in Iran. METHODS: Data were obtained from the Google Trends website. Linear regression and long short-term memory (LSTM) models were used to estimate the number of positive COVID-19 cases. All models were evaluated using 10-fold cross-validation, and root mean square error (RMSE) was used as the performance metric. RESULTS: The linear regression model predicted the incidence with an RMSE of 7.562 (SD 6.492). The most effective factors besides previous day incidence included the search frequency of handwashing, hand sanitizer, and antiseptic topics. The RMSE of the LSTM model was 27.187 (SD 20.705). CONCLUSIONS: Data mining algorithms can be employed to predict trends of outbreaks. This prediction might support policymakers and health care managers to plan and allocate health care resources accordingly.

A unique framework for the Persian clinical guidelines: addressing an evidence-based CDSS development need.

BACKGROUND AND AIM: One of the prerequisites to develop Computerised Decision Support Systems is Clinical Practice Guidelines (CPGs) which provide a systematic aid to make complex medical decisions. In order to provide an automated CPG, it is needed to have a unique structure for the CPGs. This study aims to propose a unique framework for the Persian guidelines. MATERIALS AND METHODS: 20 Persian CPGs were selected and divided into the creation and validation sets (n=10 for each). The first group was studied independently and their headings were listed; wherever possible, the headings were merged into a new heading that was applicable to all the guidelines. The developed framework was validated by the second group of the guidelines. RESULTS: Studied guidelines had a very heterogeneous structure. The number of original headings was 249; they were reduced to 14 main headings with 16 subheadings in a unique developed framework. The framework is able to represent and cover 100% of the guidelines. CONCLUSION: The heterogeneity of guidelines was high as they were not developed based on the unique framework. The proposed framework provides a layout for designing the CPGs with a homogeneous structure. Guideline developers can use this framework to develop structured CPGs. This will facilitate the integration of the guidelines into electronic medical records as well as clinical decision support systems.

Words prediction based on N-gram model for free-text entry in electronic health records.

The process of documentation is one of the most important parts of electronic health records (EHR). It is time-consuming, and up until now, available documentation procedures have not been able to overcome this type of EHR limitations. Thus, entering information into EHR still has remained a challenge. In this study, by applying the trigram language model, we presented a method to predict the next words while typing free texts. It is hypothesized that using this system may save typing time of free text. The words prediction model introduced in this research was trained and tested on the free texts regarding to colonoscopy, transesophageal echocardiogram, and anterior-cervical-decompression. Required time of typing for each of the above-mentioned reports calculated and compared with manual typing of the same words. It is revealed that 33.36% reduction in typing time and 73.53% reduction in keystroke. The designed system reduced the time of typing free text which might be an approach for EHRs improvement in terms of documentation.

Conformity of Diabetes Mobile apps with the Chronic Care Model.

BACKGROUND: Despite the growing use of mobile applications (apps) for chronic disease management, the evidence on the effectiveness of this technology on clinical and behavioural outcomes of the patients is scant. Many studies highlight the importance of the theoretical foundations of mobile-based interventions. One of the most widely accepted models for the management of chronic diseases, such as diabetes, is the Chronic Care Model (CCM). In this study, we investigated the conformity of the selected diabetes mobile apps with CCM. METHOD: We searched online journal databases related to diabetes mobile apps to find common features. Then considering the components of the CCM as a reference model, features of some popular and top-ranking apps were compared with CCM. RESULTS: Among 23 studied apps, 34 per cent of them had medium conformity and 66 per cent of these apps were in weak conformity. The self-management support component is covered by 100 per cent of them. Ninety-five per cent of apps have covered the proactive follow-up component. CONCLUSIONS: App conformance with CCM is generally weak. App developers are recommended to give greater consideration to established theoretical models in their design and implementation.

Minimum data set development for a drug poisoning registry system.

OBJECTIVE: Drug poisoning is the most prevalent type of poisoning throughout the world that can occur intentional or unintentional. Standard way for data gathering with uniform definitions is a requirement for preventing, controlling and managing of drug poisoning management. The purpose of this study was to develop a minimum data set, as an initial step, for a drug poisoning registry system in Iran. METHODS: This was descriptive and cross-sectional study that was performed in 2019. As the first step a comprehensive literature review was performed to retrieve related resources in Persian and English languages. For the second step the medical records of drug poisoning patients at Afzalipour hospital affiliated to Kerman University of Medical Sciences were assessed. Related data from these two steps were gathered by a checklist. Finally, a questionnaire that was created based on the checklist data elements and had three columns of 'essential,' 'useful, but not essential', and 'not essential' was used to reach a consensus on the data elements. Then the content validity ratio and the mean of experts' judgments were calculated for each data element. The Cronbach's alpha value for the entire questionnaire was obtained 0.9. RESULTS: The minimum data set of a drug poisoning registry system was categorised into the administrative part with three sections including 32 data elements, and clinical parts with six sections including 81 data elements. CONCLUSION: This study provides a minimum data set for development of a drug poisoning registry system. Collecting this minimum data set is critical for helping policy makers and healthcare providers to prevent, control and manage drug poisoning.

Criteria for assessing the quality of mHealth apps: a systematic review.

Objective: Review the existing studies including an assessment tool/method to assess the quality of mHealth apps; extract their criteria; and provide a classification of the collected criteria. Methods: In accordance with the PRISMA statement, a literature search was conducted in MEDLINE, EMBase, ISI and Scopus for English language citations published from January 1, 2008 to December 22, 2016 for studies including tools or methods for quality assessment of mHealth apps. Two researchers screened the titles and abstracts of all retrieved citations against the inclusion and exclusion criteria. The full text of relevant papers was then individually examined by the same researchers. A senior researcher resolved eventual disagreements and confirmed the relevance of all included papers. The authors, date of publication, subject fields of target mHealth apps, development method, and assessment criteria were extracted from each paper. The extracted assessment criteria were then reviewed, compared, and classified by an expert panel of two medical informatics specialists and two health information management specialists. Results: Twenty-three papers were included in the review. Thirty-eight main classes of assessment criteria were identified. These were reorganized by expert panel into 7 main classes (Design, Information/Content, Usability, Functionality, Ethical Issues, Security and Privacy, and User-perceived value) with 37 sub-classes of criteria. Conclusions: There is a wide heterogeneity in assessment criteria for mHealth apps. It is necessary to define the exact meanings and degree of distinctness of each criterion. This will help to improve the existing tools and may lead to achieve a better comprehensive mHealth app assessment tool.

Risk factors of Non-fatal Unintentional Home Injuries among Children under 5 Years Old; a Population-Based Study.

INTRODUCTION: In addition to the annual mortality rate, unintentional home injury may result in temporary or permanent disability and requires medical attention and continuous care in millions of children. This study aimed to explore features and risk factors of these injuries. METHODS: In this cross-sectional study, demographic variables and epidemiologic pattern of home injuries among children under 5 years of age were collected via a population-based survey in seven main cities of Khuzestan province, southwest Iran, during September 2011 to December 2012. Developing a risk stratification model, independent risk factors of unintentional home injury were determined and put to multivariate logistic regression analysis. RESULT: 2693 children with the mean age of 27.36 ± 15.55 months (1 to 60) were evaluated (50.9% boy). 827 (30.7%) cases had a history of at least one home injury occurrence since birth to study time. The most common injury mechanisms were burning with 291 (38.4%) cases, falling with 214 (28.3%) and poisoning with 66 (8.7%) cases, respectively. The independent risk factors of unintentional home injury were age ≥ 24 month (p<0.001), residency in Ahvaz city (p<0.001), mother's illiteracy (p<0.014), ethnicity (p<0.001), private housing (p=0.01), birth weight (p<0.001), and being the first child (p=0. 01). Sensitivity, specificity, and area under the ROC curve of the model designed by multivariate analysis were 53.5%, 84.8%, and 0.75 (95% CI: 0.73- 0.77; P < 0.001, figure 1), respectively. CONCLUSION: According to the findings of this study, 30.7% of the studied children were injured at least once since birth. Burning, falling, poisoning, swallowing objects, choking, and biting were the main home injury mechanisms. Age ≥ 24 months, being the first child, living in a private house, being a resident of Ahvaz city, and having an illiterate mother were found to be risk factors of home injury.