Smart medical report: efficient detection of common and rare diseases on common blood tests
2024

Smart Medical Report: Detecting Diseases from Blood Tests

Sample size: 1000000 publication 10 minutes Evidence: high

Author Information

Author(s): Németh Ákos, Tóth Gábor, Fülöp Péter, Paragh György, Nádró Bíborka, Karányi Zsolt, Paragh György Jr, Horváth Zsolt, Csernák Zsolt, Pintér Erzsébet, Sándor Dániel, Bagyó Gábor, Édes István, Kappelmayer János, Harangi Mariann, Daróczy Bálint

Primary Institution: University of Debrecen

Hypothesis

Can an AI-assisted diagnostic tool improve the detection of common and rare diseases using routine blood tests?

Conclusion

The AI tool can detect diseases significantly earlier than traditional methods, improving patient outcomes and reducing healthcare costs.

Supporting Evidence

  • The AI tool can detect diseases 1-9 months earlier than traditional methods.
  • It was validated on real-world clinical data in collaboration with Synlab Hungary.
  • The model was tested on both retrospective and prospective datasets.
  • High efficacy was demonstrated across independent laboratories and hospitals.
  • The tool significantly reduces evaluation time for healthcare providers.
  • Over one million patients' data were used for training and testing the model.
  • The study shows that AI can enhance diagnostic accuracy and efficiency.
  • Routine blood tests were the only data used for disease prediction.

Takeaway

This study shows that a smart computer program can help doctors find diseases earlier by looking at regular blood tests.

Methodology

The study developed and validated an AI tool using routine blood tests to predict various diseases, tested on over one million patients.

Potential Biases

Potential discrepancies between AI interpretations and traditional diagnoses due to reliance solely on laboratory data.

Limitations

The dataset was limited to a few regional hospital sites, which may not represent broader demographics.

Participant Demographics

The study included over one million patients with a diverse age range and both genders.

Digital Object Identifier (DOI)

10.3389/fdgth.2024.1505483

Want to read the original?

Access the complete publication on the publisher's website

View Original Publication