Deep Learning Techniques for the Dermoscopic Differential Diagnosis of Benign/Malignant Melanocytic Skin Lesions: From the Past to the Present.

There has been growing scientific interest in the research field of deep learning techniques applied to skin cancer diagnosis in the last decade. Though encouraging data have been globally reported, several discrepancies have been observed in terms of study methodology, result presentations and validation in clinical settings. The present review aimed to screen the scientific literature on the application of DL techniques to dermoscopic melanoma/nevi differential diagnosis and extrapolate those original studies adequately by reporting on a DL model, comparing them among clinicians and/or another DL architecture. The second aim was to examine those studies together according to a standard set of statistical measures, and the third was to provide dermatologists with a comprehensive explanation and definition of the most used artificial intelligence (AI) terms to better/further understand the scientific literature on this topic and, in parallel, to be updated on the newest applications in the medical dermatologic field, along with a historical perspective. After screening nearly 2000 records, a subset of 54 was selected. Comparing the 20 studies reporting on convolutional neural network (CNN)/deep convolutional neural network (DCNN) models, we have a scenario of highly performant DL algorithms, especially in terms of low false positive results, with average values of accuracy (83.99%), sensitivity (77.74%), and specificity (80.61%). Looking at the comparison with diagnoses by clinicians (13 studies), the main difference relies on the specificity values, with a +15.63% increase for the CNN/DCNN models (average specificity of 84.87%) compared to humans (average specificity of 64.24%) with a 14,85% gap in average accuracy; the sensitivity values were comparable (79.77% for DL and 79.78% for humans). To obtain higher diagnostic accuracy and feasibility in clinical practice, rather than in experimental retrospective settings, future DL models should be based on a large dataset integrating dermoscopic images with relevant clinical and anamnestic data that is prospectively tested and adequately compared with physicians.

Evidence-based clinical engineering: Health information technology adverse events identification and classification with natural language processing.

The primary goal of this project is to create a framework to extract Real-World Evidence to support Health Technology Assessment, Health Technology Management, Evidence-Based Maintenance, and Post Market Surveillance (as outlined in the EU Medical Device Regulation 2017/745) of medical devices using Natural Language Processing (NLP) and Artificial Intelligence. An initial literature review on Spontaneous Reporting System databases, Health Information Technologies (HIT) fault classification, and Natural Language Processing has been conducted, from which it clearly emerges that adverse events related to HIT are increasing over time. The proposed framework uses NLP techniques and Explainable Artificial Intelligence models to automatically identify HIT-related adverse event reports. The designed model employs a pre-trained version of ClinicalBERT that has been fine-tuned and tested on 3,075 adverse event reports extracted from the FDA MAUDE database and manually labelled by experts.

Designing and developing a mobile application for indoor real-time positioning and navigation in healthcare facilities.

BACKGROUND: Navigation portable applications have largely grown during the last years. However, the majority of them works just for outdoor positioning and routing, due to their architecture based upon Global Positioning System signals. Real-Time Positioning System intended to provide position estimation inside buildings is known as Indoor Positioning System (IPS). OBJECTIVE: This paper presents an IPS implemented as a mobile application that can guide patients and visitors throughout a healthcare premise. METHODS: The proposed system exploits the geolocation capabilities offered by existing navigation frameworks for determining and displaying the user's position. A hybrid mobile application architecture has been adopted because it allows to deploy the code to multiple platforms, simplifying maintenance and upgrading. RESULTS: The developed application features two different working modes for on-site and off-site navigation, which offer both the possibility of actual navigation within the hospital, or planning a route from a list of available starting points to the desired target, without being within the navigable area. Tests have been conducted to evaluate the performance and the accuracy of the system. CONCLUSION: The proposed application aims to overcome the limitations of Global Navigation Satellite System by using magnetic fingerprinting in combination with sensor fusion simultaneously. This prevents to rely on a single technology, reducing possible system failures and increasing the scalability.