Katie Frederickson, RN, MSN, FNP-BC, a third-year medical student, is combining nearly two decades of clinical experience with cutting-edge technology to advance research in dermatology.
Originally from Naperville, Illinois, Frederickson spent 18 years working as both a Neonatal Registered Nurse and a Dermatology Nurse Practitioner before beginning her journey in medical school. Her background in patient care has shaped her passion for dermatology and fuels her goal of matching into a dermatology residency next year.
Frederickson’s current research focuses on malignant melanoma (MM), the most aggressive form of skin cancer, where early detection is critical to improving survival outcomes. As artificial intelligence becomes increasingly integrated into healthcare, her work examines how large language models (LLMs), including GPT-5.2 and Gemini, may serve as supportive tools in identifying melanoma earlier.
Despite the growing interest in using AI for self-diagnosis and clinical decision support, Frederickson notes that the effectiveness of these technologies across diverse skin tones has not been fully understood in dermatology.
Her research explores several key questions:
- The accuracy of GPT-5.2 in melanoma detection
- Its ability to differentiate between benign and malignant skin lesions
- Comparisons between dermatologic datasets
- The model’s performance across all skin tones
Early findings show that GPT-5.2 demonstrates stable melanoma detection performance across diverse skin tones within the Milk10K dataset, suggesting limited skin-tone bias in that setting. These results highlight the potential for AI tools to support more equitable diagnostic practices in dermatology, while also underscoring the importance of testing these systems against clinically realistic datasets.
Through her work, Frederickson hopes to contribute to innovations that improve early detection of melanoma and expand access to high-quality dermatologic care for all patients.
