Author ORCID Identifier

Date of Graduation


Document Type

Dissertation (PhD)

Program Affiliation

Cancer Biology

Degree Name

Doctor of Philosophy (PhD)

Advisor/Committee Chair

Clifton David Fuller, MD., PhD.

Committee Member

Amy Catherine Moreno, MD.

Committee Member

Marina Konopleva, MD., PhD.

Committee Member

Stephen Y. Lai, MD., PhD.

Committee Member

Ruitao Lin, PhD.

Committee Member

Cullen M. Taniguchi, MD., PhD.


Oral-Cavity and oropharyngeal cancers (OC/OPC) are types of head and neck cancers that are increasing in incidence domestically. Radiation therapy (RT) is crucial in OC/OPC management. Pain is a common and challenging symptom for most patients during therapy, as nearly all patients undergoing locoregional RT in OC/OPC require analgesia for acute iatrogenic pain. Moreover, about 45% of long-term survivors report chronic pain, with more than 10% exhibiting severe chronic pain. Pain control is challenging due to the multifactorial clinical, molecular, and cellular etiology of cancer/therapy pain, as well as variation in pain assessment and the non-uniform management of pain between physicians. Therefore, there is no standard care practice pathway for patient-specific pain control during RT. Acute radiation induced pain has a reproducible temporal progression, and clinical features, and genomic markers may modify this risk. Consequently, risk stratification of patients for prophylaxis of acute on-treatment pain and prevention of acute-to-chronic pain would directly improve patient quality-of-life.

Numerous artificial intelligence (AI) prediction tools have been used recently in health care for risk stratification and personalized medicine, however, these AI-based tools have not practically tested yet in acute pain prediction and opioid doses optimization during RT in OC/OPC patients.

The motivating clinical question of this research effort is to determine if it is possible to accurately predict pain levels preemptively, and thereby maximize pain relief while treating with the minimum doses of analgesics. There is currently a gap in knowledge about how to identify the predictive features associated with pain and how to build an effective and interpretable predictive tool for acute pain prediction and analgesics optimization during RT. The scientific objective of this dissertation is to fill these gaps by identifying the clinical features and germline genomic biomarkers associated with acute pain severity and exploring the AI-predictive models that leverage patient-specific data to predict acute pain and opioid doses during RT in OC/OPC.

In the first specific aim we characterized the temporal changes in pain intensity during RT and we identified the clinical features predictive of acute pain severity/duration in OC/OPC patients. A rich set of clinicodemographic and granular pain data for OC/OPC patients at baseline, during, and after RT, including weekly pain assessments during RT were used for temporal characterization of static clinical features and dynamic patient-reported symptoms (i.e., acute pain) with statistical correlations to acute pain. Novel metrics such as area-under-the-curve measures for acute pain (AUCpain) were explored for quantitative categorization of subgroups at risk for severe acute pain.

In the second specific aim we characterized the germline genetic variants, correlated with RT-related acute pain severity, in OC/OPC patients using an association study. Single nucleotide polymorphisms (SNPs) associated with pain intensity at the end of RT were investigated using germ-line genotyping of DNA extracted from whole blood or buccal mucosa of head and neck cancer (HNC) patients collected at our institutional clinics.

Finally, in the third specific aim we explored various AI-based models for patients’ risk stratification and prediction of pain intensity, total opioid doses requirement and the analgesics efficacy in OC/OPC patients during RT. Various machine learning classification models were developed and validated using multidomain features for predicting acute pain severity and total opioid doses at the end of RT.

In conclusion, the completion of the specific aims outlined in this dissertation has allowed us to provide a comprehensive and thorough examination of various predictive features associated with acute pain development after RT in OC/OPC patients. We delved into different AI-based tools aimed at facilitating early pain risk stratification to optimize pain management. The results generated from this project are poised to have a significant impact on patient outcomes and quality of life by enabling early pain risk stratification and personalized pain management in OC/OPC patients during RT. Moving forward, future research efforts should consider exploring and validating additional features, employing advanced AI models, and assessing multiple dynamic timepoints for their eventual clinical application.


Acute pain, Radiation Therapy, Oral cavity and Oropharyngeal cancers, Artificial intelligence, Machine learning, Single nucleotide polymorphisms.

Available for download on Wednesday, April 02, 2025