Towards Digital Quantification of Ploidy from Pan-Cancer Digital Pathology Slides using Deep Learning.

Abnormal DNA ploidy, found in numerous cancers, is increasingly being recognized as a contributor in driving chromosomal instability, genome evolution, and the heterogeneity that fuels cancer cell progression. Furthermore, it has been linked with poor prognosis of cancer patients. While next-generation sequencing can be used to approximate tumor ploidy, it has a high error rate for near-euploid states, a high cost and is time consuming, motivating alternative rapid quantification methods. We introduce PloiViT, a transformer-based model for tumor ploidy quantification that outperforms traditional machine learning models, enabling rapid and cost-effective quantification directly from pathology slides. We trained PloiViT on a dataset of fifteen cancer types from The Cancer Genome Atlas and validated its performance in multiple independent cohorts. Additionally, we explored the impact of self-supervised feature extraction on performance. PloiViT, using self-supervised features, achieved the lowest prediction error in multiple independent cohorts, exhibiting better generalization capabilities. Our findings demonstrate that PloiViT predicts higher ploidy values in aggressive cancer groups and patients with specific mutations, validating PloiViT potential as complementary for ploidy assessment to next-generation sequencing data. To further promote its use, we release our models as a user-friendly inference application and a Python package for easy adoption and use.

The impact of COVID-19 on patients with chronic pain seeking care at a tertiary pain clinic.

Empirical data on the health impacts of the COVID-19 pandemic remain scarce, especially among patients with chronic pain. We conducted a cross-sectional study matched by season to examine patient-reported health symptoms among patients with chronic pain pre- and post-COVID-19 pandemic onset. Survey responses were analyzed from 7535 patients during their initial visit at a tertiary pain clinic between April 2017-October 2020. Surveys included measures of pain and pain-related physical, emotional, and social function. The post-COVID-19 onset cohort included 1798 initial evaluations, and the control pre-COVID-19 cohort included 5737 initial evaluations. Patients were majority female, White/Caucasian, and middle-aged. The results indicated that pain ratings remained unchanged among patients after the pandemic onset. However, pain catastrophizing scores were elevated when COVID-19 cases peaked in July 2020. Pain interference, physical function, sleep impairment, and emotional support were improved in the post-COVID-19 cohort. Depression, anxiety, anger, and social isolation remained unchanged. Our findings provide evidence of encouraging resilience among patients seeking treatment for pain conditions in the face of the COVID-19 pandemic. However, our findings that pain catastrophizing increased when COVID-19 cases peaked in July 2020 suggests that future monitoring and consideration of the impacts of the pandemic on patients' pain is warranted.

Classifying chronic pain using multidimensional pain-agnostic symptom assessments and clustering analysis.

Chronic pain conditions present in various forms, yet all feature symptomatic impairments in physical, mental, and social domains. Rather than assessing symptoms as manifestations of illness, we used them to develop a chronic pain classification system. A cohort of real-world treatment-seeking patients completed a multidimensional patient-reported registry as part of a routine initial evaluation in a multidisciplinary academic pain clinic. We applied hierarchical clustering on a training subset of 11,448 patients using nine pain-agnostic symptoms. We then validated a three-cluster solution reflecting a graded scale of severity across all symptoms and eight independent pain-specific measures in additional subsets of 3817 and 1273 patients. Negative affect­related factors were key determinants of cluster assignment. The smallest subset included follow-up assessments that were predicted by baseline cluster assignment. Findings provide a cost-effective classification system that promises to improve clinical care and alleviate suffering by providing putative markers for personalized diagnosis and prognosis.

Acute Pain Predictors of Remote Postoperative Pain Resolution After Hand Surgery.

INTRODUCTION: Chronic postsurgical pain (CPSP) is a global issue with high prevalence. This study compared acute pain descriptors among patients undergoing carpal tunnel release (CTR) or trigger finger release (TFR). We hypothesized worst pain intensity on postoperative day (POD) 10 would be best to predict the time to pain resolution. METHODS: In this secondary analysis of a negative, randomized, double-blind placebo-controlled trial, adult veterans undergoing CTR or TFR were enrolled January 2012-January 2014, with data analysis February 2020-October 2020. Participants were randomized to receive minocycline 200 mg or placebo 2 h prior to the operation, then minocycline 100 mg or placebo twice daily for 5 days. The Brief Pain Inventory, assessed daily, captured three pain scores: average and worst pain over the past 24 h, and current pain intensity. Fifteen acute pain descriptors based on the pain scores (clusters, mean, median, pain scores on POD 10, and linear slopes) were compared as predictors of time to pain resolution. RESULTS: Of 131 randomized participants, 114 (83 CTR, 31 TFR) were included. Average pain over the last 24 h reported on POD 10 best predicted time to pain cessation. Every one-point increase in the average pain score was associated with a 36.0% reduced rate of pain cessation (HR, 0.64, 95% CI 0.55-0.74, p < 0.001). Average pain on POD 10 was significantly associated with the development of CPSP at 90 days (OR 1.74, 95% CI 1.30-2.33, p value < 0.001). The optimal cutoff score for the high-risk group was determined as average pain on POD 10 ≥ 3. CONCLUSIONS: This study validates prior work and demonstrates the importance of assessing pain severity on POD 10 to identify patients at high risk for CPSP who are most likely to benefit from early pain intervention. Future research in diverse surgical cohorts is needed to further validate pain assessment on POD 10 as a significant predictor of CPSP.

Predicting the Incidence of Pressure Ulcers in the Intensive Care Unit Using Machine Learning.

BACKGROUND: Reducing hospital-acquired pressure ulcers (PUs) in intensive care units (ICUs) has emerged as an important quality metric for health systems internationally. Limited work has been done to characterize the profile of PUs in the ICU using observational data from the electronic health record (EHR). Consequently, there are limited EHR-based prognostic tools for determining a patient's risk of PU development, with most institutions relying on nurse-calculated risk scores such as the Braden score to identify high-risk patients. METHODS AND RESULTS: Using EHR data from 50,851 admissions in a tertiary ICU (MIMIC-III), we show that the prevalence of PUs at stage 2 or above is 7.8 percent. For the 1,690 admissions where a PU was recorded on day 2 or beyond, we evaluated the prognostic value of the Braden score measured within the first 24 hours. A high-risk Braden score (<=12) had precision 0.09 and recall 0.50 for the future development of a PU. We trained a range of machine learning algorithms using demographic parameters, diagnosis codes, laboratory values and vitals available from the EHR within the first 24 hours. A weighted linear regression model showed precision 0.09 and recall 0.71 for future PU development. Classifier performance was not improved by integrating Braden score elements into the model. CONCLUSION: We demonstrate that an EHR-based model can outperform the Braden score as a screening tool for PUs. This may be a useful tool for automatic risk stratification early in an admission, helping to guide quality protocols in the ICU, including the allocation and timing of prophylactic interventions.