Facebook has developed an algorithm that can work back from a food picture to ingredients. Then it delivers the gourmand photographer the recipe. This technology can elevate ethnographic/food/beverage research to new heights — as researchers gather photos across demographic groups and locations, they can see which ingredients are emerging and establishing themselves as crowd favorites.
Building on what I had been pontificating and sharing on AI and autism-related research, I came across this break through from Princeton University. According to the report on the study, ” The method sorted among 120,000 mutations to find those that affect the behavior of genes in people with autism. Although the results do not reveal exact causes of cases of autism, they reveal thousands of possible contributors for researchers to study. ”
Here’s why and how AI saved thousands and thousands of dollars of research budget, not to mention precious time to arrive at robust findings:
“Prior to this computational achievement, the conventional way to glean such information would be painstaking laboratory experiments on each sequence and each possible mutation in that sequence. This number of possible functions and mutations is too big to contemplate — an experimental approach would require testing each mutation against more than 2,000 types of protein interactions and repeating those experiments over and over across tissues and cell types, amounting to hundreds of millions of experiments. Other research groups have sought to accelerate this discovery by applying machine learning to targeted sections of DNA, but had not achieved the ability to look at each DNA unit and each possible mutation and the effects on each of more than 2,000 regulatory interactions across the whole genome.
“What our paper really allows you to do is take all those possibilities and rank them,” said Park. “That prioritization itself is very useful, because now you can also go ahead and do the experiments in just the highest priority cases.”
Lastly, the system calibrates its predictions based on known disease-causing mutations and develops a “disease impact score,” an assessment of how likely a given mutation is to have an effect on disease.”
The sample of families who participated in the study was under n=1,800. It’s not too large, but strong enough for analytics. It goes to show us the efficacy of the method as well. The announcement also notes that the same method could be applied to cancer research and heart disease as well. This is an incredible step forward for humanity.
I hate the puzzle. As a mom and a data-driven person, I don’t understand why we cannot solve for the autism puzzle when 1 in 59 children in the US are diagnosed with autism and the diagnoses are increasing exponentially. I hear it’s so variable, so multi-factorial and if you meet one individual on the spectrum, you meet one person on the spectrum. True. But have we ever shied away for solving for multi-faceted problems for brands who were willing to pay hundreds of thousands of dollars for a single measurement project? I would argue that we do have the statistical skills and life science knowledge to solve for autism. We lack the data.
We lack the kind of robust datasets that connect the dots between all that our children are exposed to and all that they show and do. When we have these threads of data, we can test for a myriad of variances simultaneously leveraging AI driven data science platforms. While a typical scientific study may be testing 1-5 hypotheses, we can go through 100s of hypotheses with AI in one study and quickly improve upon our knowledge.
My hope is for organizations such as the NJ AutismCenter of Excellence, Albert Einstein College of Medicine, Duke University, Epidemic Answers and others (e.g., ABA agencies) who are at the frontiers of this issue to be able to pool the data we need in this field. We need to understand why our children are having sensory motor issues that lead to behaviors. We need to tease out the environment’s impact on autism. And we need to empower our practitioners to optimize on therapies (ABA, speech, OT) and alternative interventions (homeopathy, neuro feedback, acupressure, etc. ) so that our children can have happy, productive lives.
If you needed a bit of help and support, would you be more likely to talk to a robot than a live therapist? WoeBot and Wysa apps are just two apps at the forefront of AI-based therapies providing patients with cognitive behavior therapy on demand.
Wondering how the idea would resonate, I pressure tested this concept among my family and friend circle. Guess what happened: the introverts lit up at the idea! For them, opening up to a person who might layer more judgment and stress to the process (however unintentional it may be) was adding to their emotional burden. Talking to a ‘machine’ to get answers was actually opening up the path to therapy for these individuals. Of course we still need to see data on how effective the bots are vs live therapists— broken across issues and challenges. Then there is the issue of insurance coverage and process. Maybe this is a good way to side step it all!
If AI-based therapy helps people who otherwise would not have sought help, wouldn’t we be closer to finding balance and peace? What do you think?
As underscored in Axios’s Login newsletter today, Microsoft is gearing up for a series of AI initiatives. One of them is the acquisition of Lobe that enables organizations to build AI apps and interfaces without needing to know how to code. Imagine how many departments’ ideas can now come to life more easily — from employee training, customer service communication to patient monitoring. Watch this trend as AI becomes the underlying technology in b-to-c and b-to-b service.