Predictive and Causal Analysis of No-Shows for Medical Exams During COVID-19: A Case Study of Breast Imaging in a Nationwide Israeli Health Organization

"No-shows", defined as missed appointments or late cancellations, is a central problem in healthcare systems. It has appeared to intensify during the COVID-19 pandemic and the nonpharmaceutical interventions, such as closures, taken to slow its spread. No-shows interfere with patients continuous care, lead to inefficient utilization of medical resources, and increase healthcare costs. We present a comprehensive analysis of no-shows for breast imaging appointments made during 2020 in a large medical network in Israel. We applied advanced machine learning methods to provide insights into novel and known predictors. Additionally, we employed causal inference methodology to infer the effect of closures on no-shows, after accounting for confounding biases, and demonstrate the superiority of adversarial balancing over inverse probability weighting in correcting these biases. Our results imply that a patients perceived risk of cancer and the COVID-19 time-based factors are major predictors. Further, we reveal that closures impact patients over 60, but not patients undergoing advanced diagnostic examinations.

On the Role of Artificial Intelligence in Medical Imaging of COVID-19

The global COVID-19 pandemic has accelerated the development of numerous digital technologies in medicine from telemedicine to remote monitoring. Concurrently, the pandemic has resulted in huge pressures on healthcare systems. Medical imaging (MI) from chest radiographs to computed tomography and ultrasound of the thorax have played an important role in the diagnosis and management of the coronavirus infection. We conducted the, to date, largest systematic review of the literature addressing the utility of Artificial Intelligence (AI) in MI for COVID-19 management. Through keyword matching on PubMed and preprint servers, including arXiv, bioRxiv and medRxiv, 463 papers were selected for a meta-analysis, with manual reviews to assess the clinical relevance of AI solutions. Further, we evaluated the maturity of the papers based on five criteria assessing the state of the field: peer-review, patient dataset size and origin, algorithmic complexity, experimental rigor and clinical deployment. In 2020, we identified 4977 papers on MI in COVID-19, of which 872 mentioned the term AI. 2039 papers of the 4977 were specific to imaging modalities with a majority of 83.8% focusing on CT, while 10% involved CXR and 6.2% used LUS. Meanwhile, the AI literature predominantly analyzed CXR data (49.7%), with 38.7% using CT and 1.5% LUS. Only a small portion of the papers were judged as mature (2.7 %). 71.9% of AI papers centered on disease detection. This review evidences a disparity between clinicians and the AI community, both in the focus on imaging modalities and performed tasks. Therefore, in order to develop clinically relevant AI solutions, rigorously validated on large-scale patient data, we foresee a need for improved collaboration between the two communities ensuring optimal outcomes and allocation of resources. AI may aid clinicians and radiologists by providing better tools for localization and quantification of disease features and changes thereof, and, with integration of clinical data, may provide better diagnostic performance and prognostic value.