Qi Lu
March 1, 2022
Article of the Month – March 2022
March 1, 2022Oncora Medical
Overview
Oncora Medical is a digital health company that focuses on collecting and applying data in healthcare-related decisions, mainly to improve the quality of radiation therapy treatments. Oncora’s intuitive software platform was initiated in 2014 based on the collaborative work of data scientists, clinicians, machine learning experts, and software developers.
Oncora Medical’s AI and IT Aspects
In conjunction with oncologists at MD Anderson care center, Oncora developed the patient care platform to collect necessary data about oncology patients. The patient care platform can:
Improve clinical documentation and workflow: For example, a custom-designed application developed for radiation oncologists working in high-volume cancer care centers. Also, generating documentation in less time, allowing physicians to dedicate more time to patient care.
Learn from every patient: Oncora analytics can use recorded data regarding history, treatment, and outcome to create cohorts, generate detailed analyses, and infer possible outcomes. The detailed database, along with this analytic system, allows researchers to collect and use real-world information from multiple clinical sources.
In conjunction with oncologists at MD Anderson care center, Oncora developed the patient care platform to collect necessary data about oncology patients. The patient care platform can:
Improve clinical documentation and workflow: For example, a custom-designed application developed for radiation oncologists working in high-volume cancer care centers. Also, generating documentation in less time, allowing physicians to dedicate more time to patient care.
Learn from every patient: Oncora analytics can use recorded data regarding history, treatment, and outcome to create cohorts, generate detailed analyses, and infer possible outcomes. The detailed database, along with this analytic system, allows researchers to collect and use real-world information from multiple clinical sources.
Generate reliable, auditable research-grade data: Oncora’s platform allows auditable data tracking on a per-patient basis for optimum accountability.
Discover trends in real-world data: the system can identify research questions, define patient cohorts for clinical trials, and establish associations between variables like patient characteristics, treatment modalities, and multiple outcomes. (1)
Collect details about Imaging and outcomes: Oncora incorporates information obtained through validated physician and patient-reported measures, and anatomical data collected using diagnostic imaging studies. Oncora’s predictive modeling engine gathers this information in computational models that can accurately predict the outcomes of new treatments. The objective of using these predictive models is to assist doctors in designing treatments for patients decreasing the risk of side effects and increasing their effectiveness.
Collect billing and quality assurance data: Oncora patient care and analytics software allow physicians to track and organize their patients’ oncological data such as staging, performance status, and treatment plan summaries. Modules within Oncora Patient Care use validated and custom quality metrics to measure the effectiveness of care. This allows physicians to, for example, meet billing program requirements such as the new payment model for radiation oncology that was declared final by Medicaid and Medicare services.
Quantify Healthcare costs: Oncora combines treatment decisions, the total cost of care, and outcomes in one platform, which allows measuring cost-efficacy and tracking improvement. The predicted treatment response can increase prior authorization. (2)
Oncora as an Investment
Oncora has seven investors and a total funding of $5.6 million. Oncora’s lead investors include Varian Medical System, BioAdvance, and iSeed Ventures..(3)
Oncora has an established partnership with a Palo Alto-based company called Varian Medical System. The company invested 3 million dollars in collaborating on developing advanced radiation oncology tools to improve precision medicine.
Oncora’s future platform will maintain its plans with Varian to allow maximum utilization of predictive models.
Oncora has seven investors and a total funding of $5.6 million. Oncora’s lead investors include Varian Medical System, BioAdvance, and iSeed Ventures..(3)
Oncora has an established partnership with a Palo Alto-based company called Varian Medical System. The company invested 3 million dollars in collaborating on developing advanced radiation oncology tools to improve precision medicine.
Oncora’s future platform will maintain its plans with Varian to allow maximum utilization of predictive models.
Machine Learning Models Able to Predict Survival after Radiotherapy for Brain Metastasis
Oncora has led a research investigation that entails the examination of 915 patients treated with radiation therapy (RT) for brain metastases aimed to train a predictive model for overall survival at six months, one year, and three years following radiation of brain metastases. Thus, the random forest, gradient boosted decision trees, and regularized logistic regression models were trained on a subset of radiation therapy courses with 80% of the overall dataset.
After validation, results showed a 50.3% surviving status at one year, with the best-trained model attaining an area under the curve of 0.756.
Oncora has led a research investigation that entails the examination of 915 patients treated with radiation therapy (RT) for brain metastases aimed to train a predictive model for overall survival at six months, one year, and three years following radiation of brain metastases. Thus, the random forest, gradient boosted decision trees, and regularized logistic regression models were trained on a subset of radiation therapy courses with 80% of the overall dataset.
After validation, results showed a 50.3% surviving status at one year, with the best-trained model attaining an area under the curve of 0.756.
This study seeks to prove that survival after radiotherapy for brain metastasis can be predicted by machine learning model training, thus influencing treatment decisions. (4)
Machine Learning Algorithm Able to Predict Emergency Visits and Hospital Admissions in Radiation and Chemoradiation
ML models were used to record treatment data and predict emergency visits and hospitalization for cancer therapy patients. Noticeable performance was prominent for patients with brain metastases, gynecologic neoplasms, and head and neck cancer.(6)
The Use of AI to Predict Acute Radiation Toxicities for Neck and Head Patients
A group of researchers developed a machine learning system capable of predicting unplanned hospitalizations (≥3 months from radiation therapy start), feeding tube placement, and significant weight loss (≥10% from radiation therapy (RT) starts) in patients with head and neck (HN) cancer secondary to radiation therapy. The models were trained using 1896 RT courses for HN cancer. More than 700 variables were collected for each course, including demographics, tumor characteristics, prior treatment, and RT details. Afterward, the best-performing model was evaluated with 225 consecutive RT courses for HN cancer. Models with an area under the curve (AUC) > 0.70 were considered clinically valid. In the validation set, unplanned hospitalizations incidence was 14,2%, with an AUC of 0.64. The incidence of feeding tube placement was 23.1% with an AUC of 0.755, and significant weight loss occurred in 14.2% with an AUC of 0.751
Though the model of unplanned hospitalization did not reach clinical validity, further improvement of AI oncology approaches could improve medical attention by identifying patients who may benefit from early intervention concerning feeding tube placement and weight loss due to RT for HN cancer. (7)
Machine Learning Can Predict Hospitalizations After Radiation
From a decreased quality of care to higher budgets, unplanned hospitalizations represent a problem in patients receiving radiation therapy (RT). Considering that, a study was made using a machine learning approach to predict unplanned hospitalizations in patients undergoing RT for gastrointestinal (GI) cancer within 30 days of treatment. The team analyzed 836 abdominal (gastric, pancreatic, biliary, and hepatic) and 514 pelvic (rectal and anal) courses of RT for GI cancers. Multiple variables and data from unplanned hospitalization within 30 days of RT were obtained from institutional databases. The models were trained with 670 abdominal and 423 pelvic cases. After evaluation, the best model was validated on the subsequent 166 abdominal and 91 pelvic cases. Models with an area under the curve (AUC) > 0.70 were considered clinically valid.
From a decreased quality of care to higher budgets, unplanned hospitalizations represent a problem in patients receiving radiation therapy (RT). Considering that, a study was made using a machine learning approach to predict unplanned hospitalizations in patients undergoing RT for gastrointestinal (GI) cancer within 30 days of treatment. The team analyzed 836 abdominal (gastric, pancreatic, biliary, and hepatic) and 514 pelvic (rectal and anal) courses of RT for GI cancers. Multiple variables and data from unplanned hospitalization within 30 days of RT were obtained from institutional databases. The models were trained with 670 abdominal and 423 pelvic cases. After evaluation, the best model was validated on the subsequent 166 abdominal and 91 pelvic cases. Models with an area under the curve (AUC) > 0.70 were considered clinically valid.
The incidence of 30-day unplanned hospitalizations was 12.3% (13.3% abdominal cohort vs. 10.7% pelvic cohort). The best models showed an AUC of 0.82 for the abdominal cohort and 0.78 for the pelvic cohort. These results suggest that in GI cancer patients undergoing RT, a machine learning system could identify individuals at risk of 30-day unplanned hospitalization.(8)