Introduction. To reduce protocol non-compliance as a confounding variable impacting trial outcome, Pediatric Oncology Group (POG) mandated pre-radiation quality assurance review in POG 9426, a trial in pediatric early stage Hodgkin’s disease (HD). This report documents the impact of this quality assurance program. Patients and Methods. POG 9426 investigated response-based therapy in Stages IA, IIA, and IIIA1 HD without large mediastinal masses. Early complete responders to 2 cycles of ABVE received 25 Gy of radiation therapy (XRT) to involved field(s). Partial responders to 2 cycles of ABVE received 2 more cycles of ABVE before XRT. A minimum 2 cm XRT field margin was required on all imaged diseases, as a first step in the transition from historical standard XRT field design to image-based field design. Before XRT, initial and response imaging and XRT planning films were submitted for Pre-radiation Review (PR) at QARC. Treating radiation oncologists were notified within 24 hours as to whether plans were compliant or required revision. In some cases, multiple revisions were required. The 9426 Protocol Coordinators conducted a Final Review (FR) of protocol compliance at a later date. POG 9426 enrolled 294 patients, including 246 from 85 POG institutions and 48 from 33 CCG institutions. After the first 28 cases, the directorship of QARC changed. Forty-seven cases were invaluable (incomplete submission of data) and 31 patients were removed from study before XRT leaving a total of 216 patients with both PR and FR for analysis. Results. Thirty-nine of 53 (74%) cases from institutions exempt from the requirement for pre-radiation data submission and 137 of 163 (84%) cases from non-exempt institutions submitted data for PR, indicating widespread and voluntary compliance with centralized PR at Quality Assurance Review Center (QARC). Sixteen of 40 (40%) of cases not submitted for PR were judged major protocol violations at FR, compared with 23 of 176 cases (13%) subjected to PR. At PR, modifications to achieve protocol compliance were suggested in all but 40 cases. In only 19 were modifications not made, suggesting widespread willingness to change radiation field design to achieve protocol compliance. There were discrepancies between the PR and FR in 13 of the 176 cases. The causes for disparity were interpretation of “equivocal” disease (4), gross disease (5), and adequacy of margin (3), or difference in studies available for the two reviews (1). Five (39%) of the 13 disparate reviews occurred in the initial 13 of 176 (11%) reviews, suggesting a learning curve in interpreting protocol intent. Conclusions. There was widespread acceptance of the concept of centralized pre-radiation quality assurance review and willingness both to submit diagnostic, response, and radiation treatment planning images and to implement recommended changes. We believe this to be the first centralized pre-therapy review and intervention in a U.S. based cooperative trial group. Interventions were frequently required and offered an excellent opportunity for investigator education. There were fewer major protocol violations at FR in cases subjected to PR than in cases not submitted for PR, indicating a major impact on eliminating protocol non-compliance as a variable influencing outcomes in cooperative group trials.

No abstract

No abstract

No abstract

Building a usable radio monitoring automatic speech recognition (ASR) system is a challenging task for under-resourced languages and yet this is paramount in societies where radio is the main medium of public communication and discussions. Initial efforts by the United Nations in Uganda have proved how understanding the perceptions of rural people who are excluded from social media is important in national planning. However, these efforts are being challenged by the absence of transcribed speech datasets. In this paper, The Makerere Artificial Intelligence research lab releases a Luganda radio speech corpus of 155 hours. To our knowledge, this is the first publicly available radio dataset in sub-Saharan Africa. The paper describes the development of the voice corpus and presents baseline Luganda ASR performance results using Coqui STT toolkit, an open source speech recognition toolkit.