Exponential Medicine ended up posting well over 30 hours of presentations this week.
Here is a list of the Top 50 Most Interesting Ideas of the week.
The most interesting part of the conference, was comparing the progress made since 2016 when I first attended.
Those things that were just ‘ideas’ in 2016, now have multiple companies and startups working on them. Those areas that were ‘early in development’ have now received FDA approval and are being implemented.
Admittedly, in 2016 I thought there likely was a bit too much ‘hype’ surrounding what was discussed at the conference. But as time has shown, I was incorrect.
Watch post as video below, or on your podcast app under: Gregory Schmidt
1. “If AI can replace your doctor. AI should replace your doctor”
2. AI Algorithms with FDA Approval: Didn’t realize there are already 13
3. ML being used to schedule operating room booking: based on the specific patient characteristics, with results of 30% more free time in OR schedule.
4. Machine learning is accessible: I’ve been putting off getting into trying out machine learning first hand primarily because of my mathematic incompatibility. The folks from fast.ai made a compelling case that provided someone has a background with some programming, they will be able to work with some ML models. The direction of the field is becoming more plug-in-play.
…plus a bunch of other AI stuff that has been in news over last year…
Life & Health Extension
5. Full body MRI + Whole Genome Sequencing: of healthy individuals at Human Longevity results in 14% with directly actionable findings (cancers, tumours, a fib, aneurysms, etc) and 40% with findings of long term value (genetic predisposition, hepatosteatosis, etc). Estimated 1-9 years of life added to Health Nucleus clients.
6. Focus is not just on anti-aging (slowing aging), but de-aging (becoming younger).
7. Wnt pathway manipulation is in human clinical trials (stage 1, 2) with stage 3 in 2019. Remarkable results in cartilage regeneration, tendon repair, reversal of dementia.
8. Melanopsin is the blue light sensing protein responsible for circadian rhythm. Don’t mess with it.
9. Eating outside of a 10-12hr time window for rats has dramatic negative consequences on their health. Eating within a 10 hour time window reverses disease in rats.
10. Physical forces on cells, mechanobiology, impacts how the cell behaviours (eg. oncogenic or not).
Genomics & Cellular Biology
8. NIH DNA Sequencing Cost: Appears, I’ve been mis-reading the NIH Cost Per Genome chart for years. The linear down-sloping line marking Moore’s Law is in fact already following an exponential decline in price. The genomic sequencing cost line that takes a steep drop/swoop starting in 2008 actually outpaces the exponential decline in price.
9. De-extinction: retrieval of DNA from ancient animals (Moa Bird, Wooly Mammoth)
10. The value extracted from analyzing DNA: is worth more than the cost of sequencing DNA. This means anticipated massive growth for companies built on DNA analysis.
11. Custom made viruses for dogs: very easy to buy online. Next up - custom mail order viruses for humans - and the biosecurity threats that come with it.
12. By 2020: half of all newborns in China will be genetically sequenced
13. Eugenics is back in vogue.
14. Work being done in telepathy (yes, brain-to-brain and brain-to-brain-to-brain). Surprising advancements in being able to ‘read’ and ‘write’ brains.
15. We are moving beyond implanted deep brain stimulation, to optogenetics (use of light to trigger brain), to non-invasive techniques.
16. Biofeedback (from galvanic skin response) can help with epilepsy in treatment resistant cases.
17. Meditation physically alters anatomy of the brain, and also makes brain anatomically younger than the patient’s age.
18. Mary Lou Jepsen at OpenWater is working on a crazy new ‘fMRI’ grade imaging device she believes will have a price point of $1000 to buy, and can be worn on the body. This device will be a step towards being able to continuously read thoughts.
19. An electrode placed into the temporal lobe enhanced memory.
20. Vocal Biomarkers: We heard two years ago about using voice to detect mood and depression. This time, we heard about it being used to diagnose cardiac health, pneumonia, COPD, heart attack, Parkinson’s, Alzheimer’s, Dementia.
21. Multiple new markers: pupil response (will be readily available as this is built into AR/VR tech), galvanic skin response, facial thermal changes, breath CO2 composition.
Pharmaceuticals / Clinical Trials
22. 3D printed medications is already in prototype phase [YouTube, TED - Daniel Kraft].
23. BlockChain for pharmaceuticals supply chain: This was also something people talked about ‘in theory’ two years ago. Now it is happening (Mediledger Project)
24. 74% of biopharma compounds are first in class.
25. NODE.Health - goal to be The health technology validation network, with goal of being able to deploy patient facing apps directly into the part of their care pathway that is required.
26. Science 37 has create a complete clinical research trials platform built around the patient that does not require a physician clinic or hospital for trial involvement.
27. Surgical quality metrics: will be improved and monitored much closer with rise of robots in surgery and VR surgical training.
28. Within first anchoring suture thrown by a surgeon using tools connected to haptic and sensor monitors, Dr Pugh can determine the quality and competency of a surgeon.
29. Touch Surgery’s long term goal: a real time guidance system/map live in the OR. Already doing trials with this.
30. Focused Ultrasound is exploding with indications, research, and clinical trials going on for therapy in almost every body system.
Electronic Medical Records
34. Non-clinicians display medical data really well: ePatient Dave showed some really great data visualizations by @katiemccurdy and Michael Morris [YouTube @HL7].
35. Clalit won the NEJM SPRINT Data Analysis Challenge, with their calculation of an individual patient’s risk/benefit of choosing the hypertension therapy. This outperformed the ‘aggregate’ / ‘on average’ recommendations previously made in the trial. [NEJM - Winners]
36. Patient values in algorithms: patient’s values of different outcomes (eg. stroke, fall, bleed) can be weighted by the patient and incorporated directly into the recommendation model (as shown by Clalit).
37. Hospital system have all the same reasons to shun interoperability and place a moat around their data that EHR companies do. They loose money if patient’s leave their system. We can’t ONLY blame the EHR companies for this interoperability disaster.
Healthcare at Scale / Global Health
38. It is impossible to use current medical model in the rest of the world: I’ve been talking about this for years, though have not had good facts to back me up. Jeremy Howard had great slide from World Economic Forum, who showed for Nigeria to match in 2030 the OECD physician levels, they’d require spending 10x more on their public health expenditure, and it would take 300 years to train enough doctors.
39. FDA working on final touches of a new Software as a Medical Device (SaMD) fast-track.
40. WebMD has 76 million unique monthly users (74% on mobile).
41. Digital Human Avatars - are looking very convincing (aka almost creepy) this year. They almost certainly will be the triage and first contact for patients with healthcare. Also, another avatar usage, “Your future self avatars” could improve behaviour change.
42. Who is not there? when looking at any medical data, consider, which patients have been missed or not captured by the results. It likely is those patients who don’t show to clinic, or don’t have a zip code. These are the patients with the highest burden of disease, and the one’s that need the closest attention.
43. Voice controlled technologies: can be bought by consumers for $40 (eg. Alexa), this is dramatically cheaper than a phone, tablet, laptop, or gaming console making this a widely available technology. This opens many new possibilities, as the big tech companies battle over dominance in voice.
44. When you look at the numbers, telehealth has been a failure.
45. Asynchronous communication may be as valuable as the printing press. And transform healthcare.
46. 95% of urgent care and primary health care can be done without in-person visits (per Jay Parkinson)…. In the past, surgeons were hesitant on anesthesia, because then they couldn’t hear the cries of their patient to help guide surgeries. Change in medicine takes time.
47. When companies are ranked by their Exponential Qualities, those with the highest rating outperform the S&P by several fold.
48. Alternative payment models (APM) (eg pay-for-performance, etc) are shifting cost/risk to the provider, and this appears to be working in getting providers to become smart about data, tracking their outcomes, and working to become better.
49. We use a ‘push model’ for education. We need to move to a ‘pull model’.
50. Watch out for quantum computing. Lots of $$ moving in to space. A lot of momentum in last 1-2 years.
VK Bonus. Vinod Khosla remains on track, with his prediction that 80% of what physicians do today will be done by machine. 'When I say machines will do 80% of what doctors do, people get upset. When I say nurses get upscaled to do more than a primary care physician, people get happy”
You may be interested in the original articles these Top 50 are based on
- Exponential Medicine 2018 Nov 4 - Notes from Day 1
- Exponential Medicine 2018 Nov 5 - Notes from Day 2
- Exponential Medicine 2018 Nov 6 - Notes from Day 3
- Exponential Medicine 2018 Nov 7 - Notes from Day 4