At MUSC Hollings Cancer Center in Charleston, South Carolina, researchers led by Haluk Damgacioglu have built something that can feel almost prophetic: a machine learning tool that predicts which cancer patients will face financial catastrophe before it happens. The study, published in JNCI Cancer Spectrum, represents a quiet but powerful shift in how oncology tackles one of cancer's hidden crises—the financial toxicity that claims as many patients' quality of life as the disease itself.
Nearly a quarter of Americans with cancer experience financial toxicity: the crushing weight of medical debt, unpaid bills, lost income, transportation costs, and the corrosive psychological stress of watching treatment bills mount. Damgacioglu describes the landscape plainly: cancer treatment is expensive, and financial strain doesn't just hurt the wallet. It fractures adherence to treatment. It delays care. It compromises survivorship. For some patients, the numbers become so daunting they stop treatment altogether.
The challenge, until now, has been timing. Researchers have long understood which populations are most vulnerable—younger patients, lower-income households, those with poor health status—but predicting individual risk has remained elusive. "There's the material side, like debt or unpaid bills, but there's also the psychological side," Damgacioglu explains. "Even worrying about how you'll pay for treatment can become a major source of stress." Early identification could change everything. Connect a patient with financial counseling and navigation resources before the crisis hits, and you don't just ease suffering—you potentially save lives.
To develop their tool, the research team analyzed national survey data from nearly 800 cancer patients undergoing or recently completing treatment. Patients were classified as experiencing financial toxicity if they reported material hardship or psychological stress: borrowing money, inability to pay medical bills, debt, bankruptcy, or anxiety about future costs. The team tested six different machine learning models, each trained on patient demographics, clinical data, and financial information. The goal was ruthless clarity—identify as many at-risk patients as possible, even at the cost of some false positives. "We didn't want to miss anyone who may experience financial toxicity," Damgacioglu said.
The winning model achieved 84% sensitivity and 75% specificity, meaning it reliably flagged patients at genuine risk while minimizing unnecessary false alarms. Using interpretable machine learning, the team uncovered the strongest predictors: younger age, lower income, poorer overall health, active cancer treatment, and higher out-of-pocket medical expenses. These aren't surprising factors individually, but together they paint a precise portrait of vulnerability.
The team translated their research into a publicly available web-based risk calculator that clinicians can use to estimate a patient's likelihood of financial toxicity and categorize it as low, moderate, or high. At Hollings itself, financial counselors who specialize in cancer care wait on the other side of that prediction—ready to connect flagged patients with navigation resources before financial stress reshapes their treatment journey or their lives.
It's a small tool born from big data, but in oncology, early intervention is everything. By naming financial toxicity before it happens, these researchers have created a bridge between prediction and prevention—a way to catch patients before they fall.