AI Personalized / Precision Medicine

MEETING PERSONALIZED MEDICINE WITH STEM CELLS

Today I will talk about my view on personalized medicine, stem cells and CRISPR. Why do drugs work on some patients and fail in other patients? Even the same disease can be different in different patients. This calls for both disease and patient tailored therapy, which we call precision medicine or personalized medicine.

IMMUNOLOGY OF CANCER STEM CELLS

Stem cells combine two properties in one cell. An adult cell can artificially reprogram to become pluripotent. We can use this trick to make patient and disease specific stem cells for dark screening.

NEUROPATHIC PAIN UNDER CRISPR

There are even people who don't have pain. They have mutations in iron channels and no pain. Neuropathic pain comes with mutations. CRISPR gene editing can be used to correct these mutations.

NEUROPATHIC PAIN: GENETIC MUTATIONS IN THE PATIENT

Study with a cohort of patients with neuropathic pain in the hands and feet. Patients receive a smartwatch and a smartphone. Blood sample is used to prepare DNA for genome sequencing. We suspect that these patients have mutations, but we don't know where. What we also need to address is upscaling.

Sentiment Analysis

13 positive sentences, and 21 negative ones were identified.

https://www.youtube.com/watch?v=S72vKQEezN4

Stem Cell Factory: http://www.stemcellfactory.de/en/technologies/index.html 

WHY HEALTHCARE COSTS ARE SO HIGH

Drew Calvert, a high school teacher from Texas, suffered a heart attack in 2017. He had insurance, yet his hospital bill totaled more than $100,000. High cost and complexity of care are endemic in our healthcare system.

HEALTHCARE TRANSPARENCY: THE REAL COST OF CARE

The government is now mandating insurance companies and hospitals to release their prices. There are wide differences in pricing across cities in America, even for common procedures. Cascade has launched its conversational AI Agent, powered by Healthcare Transparency.

Sentiment Analysis

14 positive sentences and 14 negative ones were identified.

https://www.youtube.com/watch?v=U8Rm1eZpGNQ

Cascade Health AI: https://www.cascade.health/ 

Cascade Health: https://www.cascadehealth.ai/

Advance Medicine & Research With AI

AI is creating new possibilities in healthcare, elevating the quality of care providers can deliver and giving the gift of time back to physicians to connect with patients. 

AI IN MEDICAL DIAGNOSIS

The whole idea here is to get into the AI world in medicine. We got some big problems here with diagnostic errors in this country. What we need is accuracy medicine, accurate medicine. And the most far reaching of which is to restore the patient doctor relationship.

MACHINE LEARNING TO PREDICT THE HEART DISEASE

The most powerful way to convey the power of a neural network for discerning things that human eyes can't see is the retina. Now we are seeing the retina as a gateway to kidney disease, to the control of diabetes and Alzheimer's disease. This is using imagination with neural networks training data.

THIS AI PROGRAM TO TREAT BABIES IN 13 HOURS

A program in San Diego uses multiple AI tools to take sick, neonates and babies, a sample of blood to do whole genome sequencing with interpretation and management in now 13 hours. We have to be on guard for all data sources as to embedded bias.

WILL AI ASSIST YOU WITH YOUR HEART?

People have AI power. That's what we're going to see more and more about ways to give autonomy to support patients. Patients can now do the same urine test at home. Five of them are currently in the FDA queue for 510K clearance in the UK.

VINOD'S VIRTUAL MEDICAL COACH

The virtual medical coach concept takes all of your data, all of the relevant medical literature and every layer of data. My favorite tool of technology is the smartphone ultrasound. Did AI help us in the pandemic? Short answer no.

THE FUTURE OF MEDICAL CARE

The number one thing that AI was felt to transform is time with the patient. 1 minute liberation from keyboard 1 minute. Modern medical practice is a petri dish for medical error. Machines are going to keep getting smarter. We need to get more humane.

A MESSAGE FROM FRANCIS PEABODY

So I just want to leave you with this six page paper from Francis Peabody, 1927. I encourage everyone to read this paper if they're involved in any way as a patient or as a doctor or a nurse. One of the essential qualities of the clinician is interest in humanity.

IMMORTAL: ADVANCING ML IN MEDICINE

What areas in the medical space have seen benefit from ML applications? What areas, if any, do you feel are critically underrepresented right now and could benefit from further research? I think it's unlimited. I don't know of any area that shouldn't be worked over.

THE BIGGEST BOTTLENECK IN AI FOR MEDICINE

The quality of the training data and also the labels, the labeling of the data. That's the biggest bottleneck right now because we're relying on supervised learning. There's only limited data sets that are large and well annotated. So that's forcing the need to go to self supervised learning.

MACHINE LEARNING IN MEDICAL STUDENT TRAINING

Should medical students get some AI experience learning? And the basics of CS, I think, yeah. If they understand, they're going to have such a bigger edge over the wider physician workforce. Well, it's not too late to take some classes.

HEART DISEASE AND THE SMARTPHONE SCANNER

Every patient I see in clinic, I do a smartphone echo. The images we've published multiple papers, they're as good. They're in parity now with the conventional ultrasound and echo machines. But the accuracy is not the AI at the moment. It's still human eyes that are largely responsible for the image.

Sentiment Analysis

178 positive sentences, and 107 negative ones were identified.

Common pitfalls and recommendations for using machine learning to detect and prognosticate for COVID-19 using chest radiographs and CT scans: https://www.nature.com/articles/s42256-021-00307-0 

Preparing the healthcare workforce to deliver the digital future: https://topol.hee.nhs.uk/wp-content/uploads/HEE-Topol-Review-2019.pdf