Trust-Augmented Deep Reinforcement Learning for Federated Learning Client Selection.

In the context of distributed machine learning, the concept of federated learning (FL) has emerged as a solution to the privacy concerns that users have about sharing their own data with a third-party server. FL allows a group of users (often referred to as clients) to locally train a single machine learning model on their devices without sharing their raw data. One of the main challenges in FL is how to select the most appropriate clients to participate in the training of a certain task. In this paper, we address this challenge and propose a trust-based deep reinforcement learning approach to select the most adequate clients in terms of resource consumption and training time. On top of the client selection mechanism, we embed a transfer learning approach to handle the scarcity of data in some regions and compensate potential lack of learning at some servers. We apply our solution in the healthcare domain in a COVID-19 detection scenario over IoT devices. In the considered scenario, edge servers collaborate with IoT devices to train a COVID-19 detection model using FL without having to share any raw confidential data. Experiments conducted on a real-world COVID-19 dataset reveal that our solution achieves a good trade-off between detection accuracy and model execution time compared to existing approaches.

Resistance to imatinib in patients with chronic myelogenous leukemia and the splice variant BCR-ABL1(35INS).

PURPOSE: In patients with chronic myelogenous leukemia (CML), point mutations in the BCR-ABL1 kinase domain are the most common cause of treatment failure with a tyrosine kinase inhibitor (TKI). It is not clear whether the splice variant BCR-ABL1(35INS) is also associated with treatment failure. PATIENTS AND METHODS: We reviewed all CML patients who had BCR-ABL1 kinase mutation analysis performed between August 1, 2007, and January 15, 2014. Patients who had BCR-ABL1(35INS) detected had their medical records reviewed to determine response to TKI therapy. RESULTS: Two hundred and eighty four patients had kinase mutation testing performed; of these, 64 patients (23%) had BCR-ABL1(35INS) detected. Forty-five patients were in chronic phase (70%), 10 were in accelerated phase (16%), 6 were in blastic phase (9%), and 3 were in other settings (5%). Of the 34 chronic phase patients who began therapy with imatinib, 23 patients (68%) failed therapy: 8 patients (24%) had primary refractory disease, 11 patients (32%) progressed, and 4 patients (12%) had disease progression after dose interruption. In contrast to the patients with disease progression or lack of response, none of 23 patients who were responding to imatinib had BCR-ABL1(35INS) detected. DNA sequencing of commonly mutated spliceosomal genes SF3B1, U2AF1, SRSF2, ZRSR2, SFA31, PRPF408, U2A565, and SF1 did not reveal mutations in seven BCR-ABL1(35INS) -positive patients tested. CONCLUSIONS: The splice variant BCR-ABL1(35INS) is frequently found in patients who are resistant to imatinib. Mutations in the commonly mutated spliceosomal proteins do not contribute to this association.