AI Diagnosis: Can AI outperform doctors?

IMAGE CREDIT:
Image credit
iStock

AI Diagnosis: Can AI outperform doctors?

AI Diagnosis: Can AI outperform doctors?

Subheading text
Medical artificial intelligence can outperform human physicians in diagnostic tasks, raising the probability of doctorless diagnosis in the future.
    • Author:
    • Author name
      Quantumrun Foresight
    • March 8, 2022

    Post text

    Some predict that medical artificial intelligence (AI) systems will pervade 90 percent of medical facilities and replace around 80 percent of human doctors’ tasks by 2025. These suggestions lead to questions about whether AI-based medical systems will eventually replace physicians in some specializations or augment their roles without actually replacing them. But for this to happen, the health care system will first have to make patients trust AI.

    AI diagnosis context

    Medical artificial intelligence (AI) can perform with accuracy and deliver cost-efficient care at scale. For instance, smartphone apps now detect skin cancer with accuracy. Algorithms identify eye diseases just as well as medical specialists. IBM’s Watson diagnoses heart disease better than many cardiologists.

    Artificial intelligence can spot subtle patterns that human doctors can easily miss. For example, a neuropathologist named, Matija Snuderl, fed the results of full-genome methylation analysis of a young girl (a former cancer patient) to an AI model to let the system classify her recurrent tumor.

    The AI analysis stated that the tumor came back as a glioblastoma, a completely different type than the pathology result. This new tumor seemed to be the outcome of radiation used to destroy previous cancer and called for another drug and radiation treatment plan. Treatment for a wrong cancer type could have ill effects on the patient without actually killing cancer. On this, Snuderl said, “If I had finalized the case just on pathology, I would have been terribly wrong.”

    Disruptive impact

    With machine learning capable of functioning on another level to human beings in terms of raw computational power, where does that leave physicians in the medical diagnostic industry? In reality, while artificial intelligence will continue to evolve and will certainly play a crucial role in visual analysis, it is more likely to envision a time when doctors will actively work alongside these tools and use AI diagnosis as a ‘second opinion’ to a human diagnosis. 

    A study found that patients are reluctant to use medical AI even when it outperforms doctors. This aversion is due to patients’ belief that their medical needs are unique and cannot be addressed by algorithms. To realize the benefits of medical AI, care providers need to discover methods to overcome patients’ resistance to Al.

    Applications for AI diagnosis

    Adoption of AI diagnosis can: 

    • Improve healthcare productivity metrics.  
    • Impact robotic surgery outcomes, allowing surgeons to treat patients with precision and decreased blood loss. 
    • Help diagnose early-stage dementia or other diseases more reliably. 
    • Make medical practice more comprehensive, accurate, and less expensive over time by reducing harmful side effects and the need for unnecessary tests.

    Questions to comment on

    • Will medical AI completely replace the roles of physicians? 
    • Can AI-based systems help in reducing medical expenses? 
    • Where will the rise of artificial intelligence leave human diagnosticians in the future?

    Insight references

    The following popular and institutional links were referenced for this insight: