ABSTRACT
Background: The impact of COVID-19 on cancer patients may be attributed not only to its direct effects on the immune system but also to delays in diagnosis and treatment. Data on the effects of COVID-19 on pancreatic ductal adenocarcinoma (PDAC) patients are scarce. Therefore, we set out to determine the impact of the pandemic on diagnosis and treatment initiation. We hypothesized that time from diagnosis to treatment would be increased in the COVID era compared to the pre-COVID era. Methods: We conducted an IRB-approved retrospective chart review of 488 patients diagnosed with PDAC from March 2019 to September 2020 at two academic medical centers. Patients were divided into two groups, based on the date of initial pathologic diagnosis. We defined the pre-COVID era as March 2019 to March 2020, the 12-month time period before California's statewide lockdown. The COVID era was defined as the 6 months following the lockdown, March 2020 to September 2020. Demographics, clinical stage, and treatment type were recorded. In addition, initial clinical encounter date, pathologic diagnosis date, and initial treatment date were also collected. All data were gathered at two large-scale academic institutions. Descriptive statistics were used in the analysis. Results: There were 333 patients diagnosed during the pre-COVID era and 155 patients during the COVID era. While race/ethnicity and age at diagnosis were statistically similar for both groups, females made up a significantly larger proportion of COVID era patients than preCOVID era patients (p= 0.02). There was no significant difference in clinical stage at diagnosis between the two groups (p= 0.84). In the pre-COVID era, 19.5% of cases were resectable, 11.1% borderline resectable, 20.1% locally advanced, and 31.8% metastatic. In the COVID era, 17.4% of patients were resectable, 11% borderline resectable, 23.9% locally advanced, and 32.9% metastatic. Median time from pathologic diagnosis to initiation of treatment was 32 days for the pre-COVID era patients and 28 days for the COVID era patients (p= 0.38). Initial treatment type was also similar between the two groups (p= 0.29). Conclusions: Fortunately, our data indicate that the COVID-19 pandemic has not significantly prevented PDAC patients from seeking care. Additionally, it does not appear that COVID-19 has delayed treatment initiation or changed initial treatment type. We believe that the successful adoption of telemedicine and other safety protocols have allowed patients with PDAC to continue receiving appropriate care during the pandemic.
ABSTRACT
Knowledge bases (KBs) and text often contain complementary knowledge: KBs store structured knowledge that can support longrange reasoning, while text stores more comprehensive and timely knowledge in an unstructured way. Separately embedding the individual knowledge sources into vector spaces has demonstrated tremendous successes in encoding the respective knowledge, but how to jointly embed and reason with both knowledge sources to fully leverage the complementary information is still largely an open problem. We conduct a large-scale, systematic investigation of aligning KB and text embeddings for joint reasoning. We set up a novel evaluation framework with two evaluation tasks, few-shot link prediction and analogical reasoning, and evaluate an array of KB-text embedding alignment methods. We also demonstrate how such alignment can infuse textual information into KB embeddings for more accurate link prediction on emerging entities and events, using COVID-19 as a case study.(1)