Empowering healthcare professionals with no-code artificial intelligence platforms for model development, a practical...

Area of Science:

• Medical Informatics
• Computational Pathology
• Artificial Intelligence in Medicine

Background:

• Artificial intelligence (AI) and machine learning (ML) offer transformative potential in healthcare diagnostics and patient management.
• A significant barrier to AI adoption in medicine is the gap between AI expertise and clinical/coding knowledge.
• No-code AI platforms are emerging to bridge this gap, enabling medical professionals to develop AI models without programming skills.

Purpose of the Study:

• To evaluate the efficacy of a no-code AI platform (Teachable Machine™) for classifying white blood cells (WBCs).
• To assess the feasibility of using accessible AI tools for hematopathology applications.
• To demonstrate the value of no-code AI for healthcare professionals lacking coding expertise.

Main Methods:

• Utilized Teachable Machine™, a no-code AI platform, to classify five common types of white blood cells.
• Employed publicly available datasets for model training and fine-tuned hyperparameters to optimize performance.
• Evaluated model performance using sensitivity, precision, and F1 scores, with validation on independent datasets.

Main Results:

• The no-code AI model achieved 97% accuracy in classifying white blood cells.
• The model demonstrated high sensitivity and precision, indicating robust performance.
• Independent validation confirmed the model's potential for further development and application in hematopathology.

Conclusions:

• No-code AI platforms like Teachable Machine™ are effective tools for medical professionals to create practical AI models in hematopathology.
• This approach democratizes AI development, allowing hands-on experience for healthcare professionals without coding prerequisites.
• The study highlights a viable pathway for integrating AI into clinical practice, starting with accessible tools and authentic datasets.