BMC Medical Informatics and Decision Making

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. 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. 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). 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.

Emerging eHealth tools could facilitate the delivery of comprehensive care in time-constrained clinical settings. One such tool is interactive computer-assisted health-risk assessments (HRA), which may improve provider-patient communication at the point of care, particularly for psychosocial health concerns, which remain under-detected in clinical encounters. The research team explored the perspectives of healthcare providers representing a variety of disciplines (physicians, nurses, social workers, allied staff) regarding the factors required for implementation of an interactive HRA on psychosocial health. The research team employed a semi-qualitative participatory method known as Concept Mapping, which involved three distinct phases. First, in face-to-face and online brainstorming sessions, participants responded to an open-ended central question: “What factors should be in place within your clinical setting to support an effective computer-assisted screening tool for psychosocial risks?” The brainstormed items were consolidated by the research team. Then, in face-to-face and online sorting sessions, participants grouped the items thematically as ‘it made sense to them’. Participants also rated each item on a 5-point scale for its ‘importance’ and ‘action feasibility’ over the ensuing six month period. The sorted and rated data was analyzed using multidimensional scaling and hierarchical cluster analyses which produced visual maps. In the third and final phase, the face-to-face Interpretation sessions, the concept maps were discussed and illuminated by participants collectively. Overall, 54 providers participated (emergency care 48%; primary care 52%). Participants brainstormed 196 items thought to be necessary for the implementation of an interactive HRA emphasizing psychosocial health. These were consolidated by the research team into 85 items. After sorting and rating, cluster analysis revealed a concept map with a seven-cluster solution: 1) the HRA’s equitable availability; 2) the HRA’s ease of use and appropriateness; 3) the content of the HRA survey; 4) patient confidentiality and choice; 5) patient comfort through humanistic touch; 6) professional development, care and workload; and 7) clinical management protocol. Drawing insight from the theoretical lens of Sociotechnical theory, the seven clusters of factors required for HRA implementation could be read as belonging to three overarching aspects : Technical (cluster 1, 2 and 3), Social-Patient (cluster 4 and 5), and Social-Provider (cluster 6 and 7). Participants rated every one of the clusters as important, with mean scores from 4.0 to 4.5. Their scores for feasibility were somewhat lower, ranging from 3.4 to. 4.3. Comparing the scores for importance and feasibility, a significant difference was found for one cluster from each region (cluster 2, 5, 6). The cluster on professional development, care and workload was perceived as especially challenging in emergency department settings, and possible reasons were discussed in the interpretation sessions. A number of intertwined multilevel factors emerged as important for the implementation of a computer-assisted, interactive HRA with a focus on psychosocial health. Future developments in this area could benefit from systems thinking and insights from theoretical perspectives, such as sociotechnical system theory for joint optimization and responsible autonomy, with emphasis on both the technical and social aspects of HRA implementation.

Statistical information (e.g., on long-term survival or side effects) may be valuable for healthcare providers to share with their patients to facilitate shared decision making on treatment options. In this pre-registered study, we assessed cancer survivors’ need for generic (population-based) versus personalized (tailored towards patient/tumor characteristics) statistical information after their diagnosis. We examined how information coping style, subjective numeracy, and anxiety levels of survivors relate to these needs and identified statistical need profiles. Additionally, we qualitatively explored survivors’ considerations for (not) wanting statistical information. Cancer survivors’ need for statistics regarding incidence, survival, recurrence, side effects and quality of life were assessed with an online questionnaire. For each of these topics, survivors were asked to think back to their first cancer diagnosis and to indicate their need for generic and personalized statistics on a 4-point scale (‘not at all’- ‘very much’). Associations between information coping style, subjective numeracy, and anxiety with need for generic and personalized statistics were examined with Pearson’s correlations. Statistical need profiles were identified using latent class analysis. Considerations for (not) wanting statistics were analyzed qualitatively. Overall, cancer survivors (n = 174) had a higher need for personalized than for generic statistics (p < .001, d = 0.74). Need for personalized statistics was associated with higher subjective numeracy (r = .29) and an information-seeking coping style (r = .41). Three statistical need profiles were identified (1) a strong need for both generic and personalized statistics (34%), (2) a stronger need for personalized than for generic statistics (55%), and (3) a little need for both generic and personalized statistics (11%). Considerations for wanting personalized cancer statistics ranged from feelings of being in control to making better informed decisions about treatment. Considerations for not wanting statistics related to negative experience with statistics and to the unpredictability of future events for individual patients. In light of the increased possibilities for using personalized statistics in clinical practice and decision aids, it appears that most cancer survivors want personalized statistical information during treatment decision-making. Subjective numeracy and information coping style seem important factors influencing this need. We encourage further development and implementation of data-driven personalized decision support technologies in oncological care to support patients in treatment decision making.

Clinical trials are mandatory protocols describing medical research on humans and among the most valuable sources of medical practice evidence. Searching for trials relevant to some query is laborious due to the immense number of existing protocols. Apart from search, writing new trials includes composing detailed eligibility criteria, which might be time-consuming, especially for new researchers. In this paper we present ASCOT, an efficient search application customised for clinical trials. ASCOT uses text mining and data mining methods to enrich clinical trials with metadata, that in turn serve as effective tools to narrow down search. In addition, ASCOT integrates a component for recommending eligibility criteria based on a set of selected protocols.

Peru has a concentrated HIV epidemic with an estimated 76,000 people living with HIV (PLHIV). Access to highly active antiretroviral therapy (HAART) expanded between 2004-2006 and the Peruvian National Institute of Health was named by the Ministry of Health as the institution responsible for carrying out testing to monitor the effectiveness of HAART. However, a national public health laboratory information system did not exist. We describe the design and implementation of an e-health driven, web-based laboratory information system - NETLAB - to communicate laboratory results for monitoring HAART to laboratory personnel, health providers and PLHIV. We carried out a needs assessment of the existing public health laboratory system, which included the generation and subsequent review of flowcharts of laboratory testing processes to generate better, more efficient streamlined processes, improving them and eliminating duplications. Next, we designed NETLAB as a modular system, integrating key security functions. The system was implemented and evaluated. The three main components of the NETLAB system, registration, reporting and education, began operating in early 2007. The number of PLHIV with recorded CD4 counts and viral loads increased by 1.5 times, to reach 18,907. Publication of test results with NETLAB took an average of 1 day, compared to a pre-NETLAB average of 60 days. NETLAB reached 2,037 users, including 944 PLHIV and 1,093 health providers, during its first year and a half. The percentage of overall PLHIV and health providers who were aware of NETLAB and had a NETLAB password has also increased substantially. NETLAB is an effective laboratory management tool since it is directly integrated into the national laboratory system and streamlined existing processes at the local, regional and national levels. The system also represents the best possible source of timely laboratory information for health providers and PLHIV, allowing patients to access their own results and other helpful information about their health, extending the scope of HIV treatment beyond the health facility and providing a model for other countries to follow. The NETLAB system now includes 100 diseases of public health importance for which the Peruvian National Institute of Health and the network of public health laboratories provide testing and results.

The rapid adoption of electronic health records (EHRs) holds great promise for advancing medicine through practice-based knowledge discovery. However, the validity of EHR-based clinical research is questionable due to poor research reproducibility caused by the heterogeneity and complexity of healthcare institutions and EHR systems, the cross-disciplinary nature of the research team, and the lack of standard processes and best practices for conducting EHR-based clinical research. We developed a data abstraction framework to standardize the process for multi-site EHR-based clinical studies aiming to enhance research reproducibility. The framework was implemented for a multi-site EHR-based research project, the ESPRESSO project, with the goal to identify individuals with silent brain infarctions (SBI) at Tufts Medical Center (TMC) and Mayo Clinic. The heterogeneity of healthcare institutions, EHR systems, documentation, and process variation in case identification was assessed quantitatively and qualitatively. We discovered a significant variation in the patient populations, neuroimaging reporting, EHR systems, and abstraction processes across the two sites. The prevalence of SBI for patients over age 50 for TMC and Mayo is 7.4 and 12.5% respectively. There is a variation regarding neuroimaging reporting where TMC are lengthy, standardized and descriptive while Mayo’s reports are short and definitive with more textual variations. Furthermore, differences in the EHR system, technology infrastructure, and data collection process were identified. The implementation of the framework identified the institutional and process variations and the heterogeneity of EHRs across the sites participating in the case study. The experiment demonstrates the necessity to have a standardized process for data abstraction when conducting EHR-based clinical studies.