Malignant brain tumors pose a substantial burden on morbidity and mortality. As clinical data collection improves, along with the capacity to analyze it, novel predictive clinical tools may improve prognosis prediction. Deep learning (DL) holds promise for integrating clinical data of various modalities. A systematic review of the DL-based prognostication of gliomas was performed using the Embase (Elsevier), PubMed MEDLINE (National library of Medicine), and Scopus (Elsevier) databases, in accordance with PRISMA guidelines. All included studies focused on the prognostication of gliomas, and predicted overall survival (13 studies, 81%), overall survival as well as genotype (2 studies, 12.5%), and response to immunotherapy (1 study, 6.2%). Multimodal analyses were varied, with 6 studies (37.5%) combining MRI with clinical data; 6 studies (37.5%) integrating MRI with histologic, clinical, and biomarker data; 3 studies (18.8%) combining MRI with genomic data; and 1 study (6.2%) combining histologic imaging with clinical data. Studies that compared multimodal models to unimodal-only models demonstrated improved predictive performance. The risk of bias was mixed, most commonly due to inconsistent methodological reporting. Overall, the use of multimodal data in DL assessments of gliomas leads to a more accurate overall survival prediction. However, due to data limitations and a lack of transparency in model and code reporting, the full extent of multimodal DL as a resource for brain tumor patients has not yet been realized.

Objectives: Refractory epilepsy is a diagnosis of recurrent seizures that requires multiple resources for optimal chronic management. The disease negatively impacts the lives of affected patients and families and poses an economic burden to the health care system. This study compares hospital costs between pediatric patients treated with antiseizure medications (ASMs) only and ASMs plus vagus nerve stimulation (VNS).Methods: Patients 0–17 years of age who were diagnosed with refractory epilepsy between January 1, 2011 and December 31, 2016, were identified from the Children's Hospital Association's Pediatric Health Information System (PHIS) database. Patients treated with ASMs only or ASMs plus VNS were included in the study and were followed 1 year prior and 2 years after meeting pre‐determined criteria for refractory epilepsy. The difference‐in‐difference (DID) approach along with the two‐part model was used to compare the changes in mean hospital costs captured in the PHIS database over time between the two cohorts.Results: One thousand one hundred thirteen patients treated with ASMs plus VNS and 3471 patients treated with ASMs only were included. At a follow‐up time of 2 years, for the ASMs‐only cohort, the adjusted all‐cause and epilepsy‐related mean annual total costs increased by $14 715 (95% confidence interval [CI]: $12 375–$17 055) and $18 437 (95% CI: $15 978–$20 896), respectively. By comparison, the adjusted all‐cause and epilepsy‐related mean annual total costs of the ASMs plus VNS cohort increased by $12 838 (95% CI: $8171–$17 505) and $15 183 (95% CI: $10 253–$20 113), respectively. Compared to ASMs only, ASMs plus VNS generated a cost savings of $3254 for epilepsy‐related annual costs per year after the index date.Significance: Compared to ASMs alone, ASMs plus VNS is a treatment modality associated with lower annual hospital costs over time. Our study shows that VNS is a cost‐beneficial treatment for a national cohort of pediatric patients with refractory epilepsy.

To conduct a descriptive assessment of patterns of utilization and cost of healthcare services and pharmacotherapies among patients with drug-resistant epilepsy (DRE) before neurostimulator implantation. Using a large United States healthcare claims database, we identified all patients with DRE who were implanted with neurostimulators between January 1, 2012, and December 31, 2019. Patients without an epilepsy diagnosis on their implantation date were excluded, as were those without (1) anti-seizure medication (ASM) dispenses within 12 months of implantation date, and (2) continuous enrollment for the 24-month period before this date. Demographic and clinical characteristics were assessed over the two-year period before implantation, as were patterns of utilization and cost of healthcare services and pharmacotherapy. Care was assessed as all-cause or epilepsy-related, with the latter defined as all medical (inpatient and outpatient) care resulting in diagnoses of epilepsy and all ASM dispenses. Eight hundred sixty patients met all selection criteria. Among these patients, comorbidities were common, including depression (27%), anxiety (30%), and learning disabilities (25%). Fifty-nine percent of patients had ≥1 all-cause hospitalizations; 57% had ≥1 epilepsy-related admissions. Patients averaged 8.6 epilepsy-related visits to physicians' offices, including 5.1 neurologist visits. Mean all-cause and epilepsy-related healthcare costs during the pre-implantation period were $123,500 and $91,995, respectively; corresponding median values were $74,567 and $53,029. Median monthly all-cause healthcare costs increased by 138% during the 24-month period (from $1,042 to $2,481 in the month prior to implantation); median epilepsy-related costs, by 290% (from $383 to $1,492). The two-year period before neurostimulator implantation is a long and costly journey. Estimates likely minimize the burden experienced during this period, given that seizure frequency and severity-and corresponding impacts on quality of life-were unavailable in these data. Further research is needed to understand the clinical, economic, and psychological impact of the time between DRE onset and implantation among qualifying patients.