T he scientific, academic, medical and data science communities have come together in the face of the COVID-19 pandemic crisis to rapidly assess novel paradigms in artificial intelligence (AI) that are rapid and secure, and potentially incentivize data sharing and model training and testing without the usual privacy and data ownership hurdles of conventional collaborations 1,2 . Healthcare providers, researchers and industry have pivoted their focus to address unmet and critical clinical needs created by the crisis, with remarkable results [3][4][5][6][7][8][9] . Clinical trial recruitment has been expedited and facilitated by national regulatory bodies and an international cooperative spirit 10-12 . The data analytics and AI disciplines have always fostered open

‘Federated Learning’ (FL) is a method to train Artificial Intelligence (AI) models with data from multiple sources while maintaining anonymity of the data thus removing many barriers to data sharing. During the SARS-COV-2 pandemic, 20 institutes collaborated on a healthcare FL study to predict future oxygen requirements of infected patients using inputs of vital signs, laboratory data, and chest x-rays, constituting the “EXAM” (EMR CXR AI Model) model. EXAM achieved an average Area Under the Curve (AUC) of over 0.92, an average improvement of 16%, and a 38% increase in generalisability over local models. The FL paradigm was successfully applied to facilitate a rapid data science collaboration without data exchange, resulting in a model that generalised across heterogeneous, unharmonized datasets. This provided the broader healthcare community with a validated model to respond to COVID-19 challenges, as well as set the stage for broader use of FL in healthcare.

No abstract

Context Pituitary adenomas (PA) are often irregularly shaped, particularly posttreatment. There are no standardized radiographic criteria for assessing treatment response, substantially complicating interpretation of prospective outcome data. Existing imaging frameworks for intracranial tumors assume perfectly spherical targets and may be suboptimal. Objective To compare a three-dimensional (3D) volumetric approach against accepted surrogate measurements to assess PA posttreatment response (PTR). Design Retrospective review of patients with available pre- and postradiotherapy (RT) imaging. A neuroradiologist determined tumor sizes in one dimensional (1D) per Response Evaluation in Solid Tumors (RECIST) criteria, two dimensional (2D) per Response Assessment in Neuro-Oncology (RANO) criteria, and 3D estimates assuming a perfect sphere or perfect ellipsoid. Each tumor was manually segmented for 3D volumetric measurements. The Hakon Wadell method was used to calculate sphericity. Setting Tertiary cancer center. Patients or Other Participants Patients (n = 34, median age = 50 years; 50% male) with PA and MRI scans before and after sellar RT. Interventions Patients received sellar RT for intact or surgically resected lesions. Main Outcome Measure(s) Radiographic PTR, defined as percent tumor size change. Results Using 3D volumetrics, mean sphericity = 0.63 pre-RT and 0.60 post-RT. With all approaches, most patients had stable disease on post-RT scan. PTR for 1D, 2D, and 3D spherical measurements were moderately well correlated with 3D volumetrics (e.g., for 1D: 0.66, P < 0.0001) and were superior to 3D ellipsoid. Intraclass correlation coefficient demonstrated moderate to good reliability for 1D, 2D, and 3D sphere (P < 0.001); 3D ellipsoid was inferior (P = 0.009). 3D volumetrics identified more potential partially responding and progressive lesions. Conclusions Although PAs are irregularly shaped, 1D and 2D approaches are adequately correlated with volumetric assessment.

No abstract