Recently, I traveled to Houston to attend the AI in Health Conference organized by The Ken Kennedy Institute. Today, the healthcare industry is on the precipice of making great strides in implementing artificial intelligence (AI) techniques that improve patient outcomes, help diagnose disease, assist in preventative care and treatments, and so much more.
For example, natural language processing (NLP) could be used to translate clinical notes and extract data from patient history to help predict illness. AI can help identify new drugs and interpret thousands of images and videos to identify T-cells that kill cancer cells at an accelerated rate compared to others. It finds new links between genetic codes, saves time with administrative tasks, and drives robotic surgeries.
The opportunity is ample. AI can help doctors exhausted from long shifts find the signal from the noise in data. Battle medics can use ultrasounds to identify internal conditions without having the years of medical training they would need otherwise, all to provide life-saving interventions before arriving at a proper medical facility. Even, one day, digital images could offer preliminary identification of conditions or register severity for parents or family members - imagine understanding the type of spider that bit a child (and if it's poisonous) from the bite marks in real-time.
Yet, for AI to truly succeed in health, it must work closely with humans. By combining the strengths of human intelligence and decision-making and the power of deep learning analysis, we can save lives and improve outcomes in the medical field.
Roadblocks to AI Adoption
Adoption of AI in healthcare remains a challenge. There are privacy and data sharing concerns (Health Insurance Portability and Accountability Act (HIPAA)), which are significant barriers to any kind of data collection. Moreover, there's a tremendous asymmetry in the data available, with far more available for healthy conditions than not healthy conditions, and even less data available for many types of conditions. To add, a lot of useful health data comes from wearable devices, meaning companies like Apple and Google own much of that data, essentially locking it up in siloes.
The primary challenge then becomes how healthcare organizations can use AI to ethically collect, analyze, and apply patient data, regardless of where it resides or the form it is in. To do this right would require time (which doctors don't have enough of and staff is feeling burned out), resources (there is a general lack of staffing and funds for AI initiatives), and training (healthcare organizations need to improve their overall technical literacy).
There's a deeply human story to tell here. Doctors are overworked and aren't data scientists, but they're extremely intelligent and educated. They need explainable, simplified, interpretable AI that can help them give better diagnoses. Human-centered AI is at the heart of the medical profession and can be a technology that helps improve lives and save time and money.
By employing the right people to interpret privacy rules and regulations and training models on what to look for in real-time, speeding up the process, and shrinking models to fit on edge devices, AI can be a system designed for people. It can be safe, efficient, and secure. It can work within the current regulatory framework at speeds that can genuinely improve outcomes.
Improving healthcare, and more specifically improving health and care, is and should always be a distinctly human-centered endeavor. Technology is simply there to amplify this endeavor, empowering researchers, scientists, practitioners and even institutions to find and deliver breakthroughs that dramatically improve the health and wellness of each of us.
This is how we have built our Vian H+AI Platform across three critical aspects of maximizing the potential of AI - building trust, transparency and accountability with AI monitoring tools, enabling wider accessibility of AI at lower cost through AI optimization, and finally delivering AI directly into the hands of users including in this case doctors, nurses, radiologists, researchers, and more, through AI-based applications.
Are you at a pivotal moment in your AI journey to empower the humans behind the technology, in healthcare, or other areas? Were you at the AI in Health conference? We would love to know your thoughts on how human-centered AI is creating meaningful impact across industries.