Key Principles of Clinical Validation, Device Approval, and Insurance Coverage Decisions of Artificial Intelligence

Artificial intelligence (AI) will likely affect various fields of medicine. This article aims to explain the fundamental principles of clinical validation, device approval, and insurance coverage decisions of AI algorithms for medical diagnosis and prediction.Discrimination accuracy of AI algorithms is often evaluated with the Dice similarity coefficient, sensitivity, specificity, and traditional or free-response receiver operating characteristic curves. Calibration accuracy should also be assessed, especially for algorithms that provide probabilities to users. As current AI algorithms have limited generalizability to realworld practice, clinical validation of AI should put it to proper external testing and assisting roles. External testing could adopt diagnostic case-control or diagnostic cohort designs. A diagnostic case-control study evaluates the technical validity/ accuracy of AI while the latter tests the clinical validity/accuracy of AI in samples representing target patients in realworld clinical scenarios. Ultimate clinical validation of AI requires evaluations of its impact on patient outcomes, referred to as clinical utility, and for which randomized clinical trials are ideal. Device approval of AI is typically granted with proof of technical validity/accuracy and thus does not intend to directly indicate if AI is beneficial for patient care or if it improves patient outcomes. Neither can it categorically address the issue of limited generalizability of AI. After achieving device approval, it is up to medical professionals to determine if the approved AI algorithms are beneficial for real-world patient care. Insurance coverage decisions generally require a demonstration of clinical utility that the use of AI has improved patient outcomes.

Key principles of clinical validation, device approval, and insurance coverage decisions of artificial intelligence

Artificial intelligence (AI) will likely affect various fields of medicine. This article aims to explain the fundamental principles of clinical validation, device approval, and insurance coverage decisions of AI algorithms for medical diagnosis and prediction. Discrimination accuracy of AI algorithms is often evaluated with the Dice similarity coefficient, sensitivity, specificity, and traditional or free-response receiver operating characteristic curves. Calibration accuracy should also be assessed, especially for algorithms that provide probabilities to users. As current AI algorithms have limited generalizability to real-world practice, clinical validation of AI should put it to proper external testing and assisting roles. External testing could adopt diagnostic case-control or diagnostic cohort designs. A diagnostic case-control study evaluates the technical validity/accuracy of AI while the latter tests the clinical validity/accuracy of AI in samples representing target patients in real-world clinical scenarios. Ultimate clinical validation of AI requires evaluations of its impact on patient outcomes, referred to as clinical utility, and for which randomized clinical trials are ideal. Device approval of AI is typically granted with proof of technical validity/accuracy and thus does not intend to directly indicate if AI is beneficial for patient care or if it improves patient outcomes. Neither can it categorically address the issue of limited generalizability of AI. After achieving device approval, it is up to medical professionals to determine if the approved AI algorithms are beneficial for real-world patient care. Insurance coverage decisions generally require a demonstration of clinical utility that the use of AI has improved patient outcomes.

Surgical-tools detection based on Convolutional Neural Network in laparoscopic robot-assisted surgery.

Laparoscopic surgery, a type of minimally invasive surgery, is used in a variety of clinical surgeries because it has a faster recovery rate and causes less pain. However, in general, the robotic system used in laparoscopic surgery can cause damage to the surgical instruments, organs, or tissues during surgery due to a narrow field of view and operating space, and insufficient tactile feedback. This study proposes real-time models for the detection of surgical instruments during laparoscopic surgery by using a CNN(Convolutional Neural Network). A dataset included information of the 7 surgical tools is used for learning CNN. To track surgical instruments in real time, unified architecture of YOLO apply to the models. So as to evaluate performance of the suggested models, degree of recall and precision is calculated and compared. Finally, we achieve 72.26% mean average precision over our dataset.

Analysis of gait pattern during stair walk for improvement of gait training robot.

This paper seeks stair gait patterns which can improve effectiveness of gait training by applying to robotic locomotion therapy system. To get applicable data for stair walking function of the system, the stair walk of five subjects were measured by a motion capture system. From the acquired data, trajectories of each joint angle and relative change of the joints were calculated in the anatomical sagittal plane. Also, we were attempt to create more natural stair gait pattern by analyzing the movement of hip on the transverse plane.

Automatic detection of hemorrhage and surgical instrument in laparoscopic surgery image.

In this paper, we propose a new method for detecting hemorrhage areas and surgical instruments in robot-assisted laparoscopic surgery images. The proposed scheme utilizes CIELAB information to identify a region of interest (ROI) and segment it. Histogram equalization and Otsu's method are also adopted to compute the detection threshold. Detection is performed automatically and additional adjustment of parameters is not needed. Experiments to verify the proposed algorithm were conducted using actual robot-assisted laparoscopic surgery images. Using the proposed algorithm, the average time consumption was 0.37 s per frame for hemorrhage identification and 0.11 s for instrument detection. The sensitivities were also high enough for practical application.

An Engineering View on Megatrends in Radiology: Digitization to Quantitative Tools of Medicine

Within six months of the discovery of X-ray in 1895, the technology was used to scan the interior of the human body, paving the way for many innovations in the field of medicine, including an ultrasound device in 1950, a CT scanner in 1972, and MRI in 1980. More recent decades have witnessed developments such as digital imaging using a picture archiving and communication system, computer-aided detection/diagnosis, organ-specific workstations, and molecular, functional, and quantitative imaging. One of the latest technical breakthrough in the field of radiology has been imaging genomics and robotic interventions for biopsy and theragnosis. This review provides an engineering perspective on these developments and several other megatrends in radiology.