Cole Lemasters used ChatGPT in mid 2025 to review his wife Kiley Lemasters radiology report and original biopsy records from July 2020, uncovering an IDH1 R132H mutation that led doctors to reclassify her brain tumor from glioblastoma to a less aggressive IDH mutant astrocytoma under the 2021 World Health Organization standards. The events unfolded primarily in Missouri, beginning with Kiley’s seizure and diagnosis in the Kansas City area at age 18, followed by their wedding in a small rural Missouri church on December 18 2020, and their later life in Columbia where Cole attended the University of Missouri and Kiley pursued studies at Moberly Area Community College while working as a barista. Stable MRI scans over five years prompted Cole to question the original grim three month prognosis after her latest imaging in July 2025, the fifth anniversary of the terminal outlook that had led her to write a will and select a casket.
The process began when Cole copied the dense text of the recent radiology report into ChatGPT seeking plain language explanations of terms describing a small new enhancing focus near the resection cavity. The AI clarified that finding as a common post treatment effect and, when Cole inquired further about the original diagnosis and long term stability, pointed to the 2021 WHO update on central nervous system tumors. That update restricted the glioblastoma label to IDH wildtype tumors only and reclassified similar appearing IDH mutant cases as astrocytomas with markedly better median survival measured in years rather than months. Cole then retrieved a box of 2020 pathology documents from his mother in law and uploaded pages one by one, approximately 60 in total over an hour, until the model highlighted the positive IDH1 R132H result in the molecular genetics section of the original report.
The original 2020 diagnosis rested on histological features available at the time and the classification rules then in force, when IDH mutation testing was not always performed routinely on every sample. Pathologists and oncologists applied the standards correctly for that era, yet the mutation data already present in the records gained new significance only after the WHO revision. Kiley’s current neuro oncologist reviewed the AI surfaced finding, confirmed its validity against the updated criteria, and adjusted the care plan from short horizon crisis management to long term monitoring. The couple, still residing in the Columbia Missouri region, shifted focus toward careers, housing, and possible family plans once the prognosis improved.
In the broader health sector AI tools already assist with rapid analysis of MRI scans through deep learning models that predict IDH mutation status from imaging alone, often achieving high accuracy without immediate need for invasive biopsy in research settings. Digital pathology systems apply convolutional networks and vision transformers to histology slides to estimate molecular profiles including IDH status, glioma grade, and other markers, speeding turnaround and reducing variability among readers. Large language models like ChatGPT help clinicians and patients parse lengthy reports, cross reference guidelines, and surface overlooked details across multi year records, serving as a second set of eyes rather than a replacement for medical judgment. Ongoing research includes fully automated MRI based networks that forecast IDH mutations preoperatively, hybrid radiomics deep learning pipelines, and AI driven molecular profiling from tissue images, with multiple 2024 and 2025 studies demonstrating performance that can match or exceed certain physician groups in controlled comparisons while still requiring clinical validation.
These developments point toward integrated workflows in which AI flags potential reclassifications, prioritizes cases for molecular testing, and supports shared decision making, particularly as classification systems evolve. For patients with long stable disease under older labels, systematic review of archived molecular data against current guidelines can alter treatment intensity and life planning. The Lemasters experience illustrates practical value when AI connects existing information across time and helps families pose precise questions to their care teams, always with final confirmation by specialists. More detail on the personal timeline appears in the account at this detailed narrative, the 2021 classification shifts are outlined by the National Brain Tumor Society at this overview, and examples of AI advancing brain tumor imaging and diagnosis are available through the National Cancer Institute at this report along with OpenAI’s summary at this post.
