Thyroid Nodule Classification for Physician Decision Support Using Machine Learning-Evaluated Geometric and Morphological Features

The classification of thyroid nodules using ultrasound (US) imaging is done using the Thyroid Imaging Reporting and Data System (TIRADS) guidelines that classify nodules based on visual and textural characteristics. These are composition, shape, size, echogenicity, calcifications, margins, and vascularity. This work aims to reduce subjectivity in the current diagnostic process by using geometric and morphological (G-M) features that represent the visual characteristics of thyroid nodules to provide physicians with decision support. A total of 27 G-M features were extracted from images obtained from an open-access US thyroid nodule image database. 11 significant features in accordance with TIRADS were selected from this global feature set. Each feature was labeled (0 = benign and 1 = malignant) and the performance of the selected features was evaluated using machine learning (ML). G-M features together with ML resulted in the classification of thyroid nodules with a high accuracy, sensitivity and specificity. The results obtained here were compared against state-of the-art methods and perform significantly well in comparison. Furthermore, this method can act as a computer aided diagnostic (CAD) system for physicians by providing them with a validation of the TIRADS visual characteristics used for the classification of thyroid nodules in US images.

Implantable Cardioverter Defibrillator mHealth App for Physician Referrals and eHealth Education: ICD-TEACH Pilot Study (Preprint)

BACKGROUND Mobile health (mHealth) decision tools for implantable cardioverter defibrillator may increase physician knowledge and overall patient care. OBJECTIVE The goals of the ICD-TEACH pilot study were to design a smartphone app or mHealth technology with a novel physician decision support algorithm, implement a direct referral mechanism for device implantation from the app, and assess its overall usability and feasibility with physicians involved in the care of patients with an implantable cardioverter defibrillator. METHODS The initial design and development of the mHealth or smartphone app included strategic collaboration from an information technology company and key stakeholders including arrhythmia specialists (electrophysiologists), general cardiologists, and key members of the hospital administrative team. A convenience sampling method was used to recruit general internists or cardiologists that refer to our local tertiary care center. Physicians were asked to incorporate the mHealth app in daily clinical practice and avail the decision support algorithm and direct referral feature to the arrhythmia clinic. Feasibility assessment, in the form of a physician survey, was conducted after initial mHealth app use (within 3 months) addressing the physicians’ overall satisfaction with the app, compliance, and reason for noncompliance; usability assessment of the mHealth app was addressed in the physician survey for technical or hardware problems encountered while using the app and suggestions on improvement. RESULTS A total of 17 physicians agreed to participate in the pilot study with 100% poststudy survey response rate. Physicians worked in an academic practice, which included both inpatient and ambulatory care. System Usability Scale was applied with an average score of 77 including the 17 participants (>68 points is above average). Regarding the novel physician decision support algorithm for implantable cardioverter defibrillator referral, 11% (1/9) strongly agreed and 78% (7/9) agreed that the algorithm for device eligibility was easy to use. Only 1 patient was referred through the direct referral system via the mHealth app during the pilot study of 3 months. Feasibility assessment showed that 46% (5/11) strongly agreed and 55% (6/11) agreed that the mHealth app would be utilized if integrated into an electronic medical record (EMR) where data are automatically sent to the referring arrhythmia clinic. CONCLUSIONS The ICD-TEACH pilot study revealed high usability features of a physician decision support algorithm; however, we received only 1 direct referral through our app despite supportive feedback. A specific reason from our physician survey included the lack of integration into an EMR. Future studies should continue to systematically evaluate smartphone apps in cardiology to assess usability, feasibility, and strategies to integrate into daily workflow.

Parallel User Interface Design of a Clinical Decision-Support Application for Desktop and Pocket Pc Platforms

In order to increase physician acceptance and use, it is necessary for clinical information systems to better support workflow and connectivity. Towards that end, it is advantageous to develop clinical applications that support a range of platforms and mobile devices. However, differences in design/development approaches, technical limitations, and user interactivity across devices result in inconsistent features and user experiences, limiting functionality, usability, and transfer of training. In the current project, a browser-based physician decision-support and order entry prototype was developed for the Windows desktop and Pocket PC in parallel. Corresponding functionality was implemented on both platforms via an iterative, user-centered design approach that utilized components of the desktop version to create the PocketPC screens. Subsequent physician feedback demonstrated high transfer of training from the desktop version to the PocketPC. The findings from this work can be applied to other multi-platform user interface projects.