Advancing Fusion with Machine Learning Research Needs Workshop Report

Machine learning and artificial intelligence (ML/AI) methods have been used successfully in recent years to solve problems in many areas, including image recognition, unsupervised and supervised classification, game-playing, system identification and prediction, and autonomous vehicle control. Data-driven machine learning methods have also been applied to fusion energy research for over 2 decades, including significant advances in the areas of disruption prediction, surrogate model generation, and experimental planning. The advent of powerful and dedicated computers specialized for large-scale parallel computation, as well as advances in statistical inference algorithms, have greatly enhanced the capabilities of these computational approaches to extract scientific knowledge and bridge gaps between theoretical models and practical implementations. Large-scale commercial success of various ML/AI applications in recent years, including robotics, industrial processes, online image recognition, financial system prediction, and autonomous vehicles, have further demonstrated the potential for data-driven methods to produce dramatic transformations in many fields. These advances, along with the urgency of need to bridge key gaps in knowledge for design and operation of reactors such as ITER, have driven planned expansion of efforts in ML/AI within the US government and around the world. The Department of Energy (DOE) Office of Science programs in Fusion Energy Sciences (FES) and Advanced Scientific Computing Research (ASCR) have organized several activities to identify best strategies and approaches for applying ML/AI methods to fusion energy research. This paper describes the results of a joint FES/ASCR DOE-sponsored Research Needs Workshop on Advancing Fusion with Machine Learning, held April 30–May 2, 2019, in Gaithersburg, MD (full report available at https://science.osti.gov/-/media/fes/pdf/workshop-reports/FES_ASCR_Machine_Learning_Report.pdf). The workshop drew on broad representation from both FES and ASCR scientific communities, and identified seven Priority Research Opportunities (PRO’s) with high potential for advancing fusion energy. In addition to the PRO topics themselves, the workshop identified research guidelines to maximize the effectiveness of ML/AI methods in fusion energy science, which include focusing on uncertainty quantification, methods for quantifying regions of validity of models and algorithms, and applying highly integrated teams of ML/AI mathematicians, computer scientists, and fusion energy scientists with domain expertise in the relevant areas.

A study of sequential high dose cyclophosphamide and high dose carboplatin with peripheral stem-cell rescue in resistant or recurrent pediatric brain tumors

PURPOSE

To determine the maximum tolerated dose (MTD) of carboplatin with autologous hematopoietic stem-cell rescue, in children with poor-prognosis brain tumors.

PATIENTS AND METHODS

A previously determined dose of cyclophosphamide with stem-cell rescue was used as a first course. In a second course, carboplatin was given for 3 days with stem-cell rescue to 20 children. The starting dose of carboplatin was 400 mg/m2/day with increments of 75 mg/m2/day in subsequent cohorts. Toxicity and tumor response were recorded.

RESULTS

There were two grade IV toxicities at the dose level of 775 mg/m2/day. There were no toxic deaths. Thus, the MTD of carboplatin was 700 mg/m2/day for 3 days. There were six complete responses (33%, 95% confidence interval [CI], 13-59%), two partial responses (11%; 95% CI, 1-35%), four with stable diseases (22%; 95% CI, 6-48%) and six progressed (33%; 95% CI, 13-59%) out of 18 assessable. Seven of the eight responses were in primitive neuroectodermal tumors (PNETs) or Germinomas. One child with a metastatic anaplastic astrocytoma had a CR. The median duration of tumor response was 10 months (range: 1.5-87 months) with two children disease free at 66 and 87 months. Actuarial survival is 21%. Median follow-up of survivors is 35 months (range: 15-87 months).

CONCLUSION

The MTD of carboplatin with stem-cell rescue is 700 mg/m2/day for 3 days. Sequential stem-cell supported cyclophosphamide and carboplatin was tolerable in children with brain tumors and produced responses in PNETs and Germinomas.

Monosomy 7 associated with pediatric acute myeloid leukemia (AML) and myelodysplastic syndrome (MDS): successful management by allogeneic hematopoietic stem cell transplant (HSCT)

Pediatric acute myeloid leukemia (AML) or myelodysplastic syndrome (MDS) with monosomy 7 is associated with poor disease-free survival when treated by conventional chemotherapy, immunosuppression or supportive measures. Hematopoietic stem cell transplant (HSCT) may improve outcomes; however, data to support this are limited. To better understand the curative potential of HSCT in these patients, all cases of AML and MDS with monosomy 7 treated by two transplant programs (1992 to present) were reviewed. A total of 16 patients were treated, all by allogeneic HSCT. Primary diagnoses were MDS (N = 5), therapy-related MDS (N = 3), AML (N = 5) and therapy-related AML (N = 3). In all, 11 patients (69%) survive event-free at 2 years with median follow-up of 986 days (range 330-2011 days). Toxicity caused deaths of the five nonsurviving patients, four of whom were transplanted with active leukemia. Allogeneic HSCT is effective therapy for childhood AML and MDS associated with monosomy 7, particularly for patients with AML in complete remission and MDS.

Transplantation of ex vivo expanded cord blood

Umbilical cord blood (CB) from unrelated donors is increasingly used to restore hematopoiesis after myeloablative therapy. CB transplants are associated with higher rates of delayed and failed engraftment than are bone marrow transplants, particularly for adult patients. We studied the ex vivo expansion of CB in an attempt to improve time to engraftment and reduce the graft failure rate in the recipients. In this feasibility study, 37 patients (25 adults, 12 children) with hematologic malignancies (n = 34) or breast cancer (n = 3) received high-dose therapy followed by unrelated allogeneic CB transplantation. A fraction of each patient's CB allograft was CD34-selected and cultured ex vivo for 10 days prior to transplantation in defined media with stem cell factor, granulocyte colony-stimulating factor, and megakaryocyte growth and differentiation factor. The remainder of the CB graft was infused without further manipulation. Two sequential cohorts of patients were accrued to the study. The first cohort had 40% and the second cohort had 60% of their CB graft expanded. Patients received a median of 0.99 x 10(7) total nucleated cells (expanded plus unexpanded) per kilogram. The median time to engraftment of neutrophils was 28 days (range, 15-49 days) and of platelets was 106 days (range, 38-345 days). All evaluable patients who were followed for 28 days or longer achieved engraftment of neutrophils. Grade III/IV acute GVHD was documented in 40% and extensive chronic GVHD in 63% of patients. At a median follow-up of 30 months, 13 (35%) of 37 of patients survived. This study demonstrates that the CD34 selection and ex vivo expansion of CB prior to transplantation of CB is feasible. Additional accrual will be required to assess the clinical efficacy of expanded CB progenitors.