In this post, you’ll read what we’ve been up to for the last year or so. You’ll learn what we’ve created and how it might benefit you and the Parkinson’s community. We’ll talk about what people with Parkinson’s really want, and why doctors, drug companies and much of scientific research have been failing them. We’ve been thinking outside the box, and we think you’ll like it. This is not a scientific article, it is written so that non-scientists can understand it and get a sense of our progress.
Sorry for not updating you more frequently…
As scientists, we’re used to communicating our work formally, and only after we’re pretty damn sure we’re finished with something and made a substantial contribution to our field. But Annette has been nagging us to write more about our progress. After all, working on PD was her idea. She was able to find funding for the project entirely from donations. Initially, she was only interested in stopping the progression of her husband’s PD. But she realized that our efforts could help others as well. It’s her call and we respect it.
So this post is a compromise. We’re not finished, in fact this project will probably always be a work in progress. But we reached a stage where we can confidently talk about it. We’ve made a lot of progress and we are very pleased with the results so far. Annette’s husband has been our only guinea pig. I’ll let Annette talk about that if she wants. But keep in mind that his progress is anecdotal. In other words, his progress does not necessarily predict success for others. And also remember that the purpose of this project is not to test, but to make predictions. We generate some of our own training data using experimental testing, but after that it’s all predictions.
PD will be stopped by computers, not human trials with single drugs
People with Parkinson’s have one wish: to have their PD cured, or at least stopped so that they don’t lose their independence and become a burden on loved ones. They want to maintain as much of a normal life as possible. PD is a degenerative, progressive disease. It won’t wait for scientists and drug companies to find something for them. They need something NOW. Not in 10 years. TODAY.
But let’s be honest: at the current rate of progress, PD will be stopped in a few generations. Not days, not months, not years, not decades, but generations. There are a few reasons for this:
- Science is a slow, methodical process.
- Traditionally, cures only happen after a thorough understanding of a disease.
- We are very, very far away from understanding PD.
- PD is an incredibly complex disease.
- PD is a multifactorial, complex neurodegenerative disease.
- Scientists don’t even know where in the body it starts, or really how it spreads.
- It is a chain reaction fed by multiple biochemical pathways.
- A number of these pathways will need to modulated simultaneously if one has any hope of slowing or stopping PD.
- Drug companies are primarily interested in treating symptoms.
- The cynical truth is that there is good money is in treating symptoms.
- Curing the disease, or even stopping progression, eliminates customers.
- Think about it. L-dopa was discovered in the 1950’s and 1960’s. How much have things really changed?
- Drugs must be tested on animals before they are tested on people.
- Animals don’t get PD, but animals can be made to show some of the symptoms of PD by modifying them chemically or genetically.
- Because of these differences, drugs that work in animals may not work in people. History teaches us that they rarely do.
- However, animal experiments can help us predict which pathways will be important for stopping PD in people.
- Human drug trials are slow and expensive.
- They are difficult enough using one drug that worked in animals.
- But if we need to test many drug cocktails??? The combinations are endless.
- There is basically no way to test many of these combinations at any conceivable amount of funding, in a human lifetime.
But it is becoming more widely accepted that drug combinations are the only way to meaningfully impact the progression of complex, degenerative diseases like PD. For more about combinatorial therapeutics, see these examples:
- Confidence from uncertainty – A multi-target drug screening method from robust control theory
- Systematic quantitative characterization of cellular responses induced by multiple signals
- Multi-target therapeutics: when the whole is greater than the sum of the parts
- Efficient discovery of anti-inflammatory small-molecule combinations using evolutionary computing
- Adaptive reference update (ARU) algorithm. A stochastic search algorithm for efficient optimization of multi-drug cocktails
- Drug Combination Studies and Their Synergy Quantification Using the Chou-Talalay Method
- Statistical Metamodeling for Revealing Synergistic Antimicrobial Interactions
- Neighbor communities in drug combination networks characterize synergistic effect
- A Diverse Stochastic Search Algorithm for Combination Therapeutics
Because of the number of combinations of possible drugs that need to interact with an unknown combination of pathways, we don’t think that traditional clinical trial paradigm for drug creation will be successful with PD. At least not within the lifetimes of anybody currently living with PD. The only hope for helping people with PD today, beyond drugs to reduce symptoms and other forms of temporary relief, is by using very smart computers. How smart?
Dr. Ben and I having been using these smart computers for years now. We’ve created our own, and created the software that they run. Before Annette found us we were already tinkering with smart computers to predict drug combinations for other diseases. It is a very cutting edge field and there’s no reason it can’t be used to help people with Parkinson’s too.
But I Want a Cure NOW, Daddy!
OK, so drug combinations are the way to go, and computers are the only way to do it. But the problem is that drug combinations are just like any new drugs. They must be tested to show they are safe and effective. This will take a decade per combination, at least!. To test them all in parallel would be impossibly difficult and expensive. It would take a Willy Wonka to make that happen.
Fortunately, we have a solution that will allow people with PD to actually use these predicted drug combinations now. The solution is to use only existing chemicals that are available over-the-counter (OTC). In fact, we think that all of the substances in our predicted combinations can be adequately selected from a library consisting of many thousands of safe, natural, mostly plant-based substances. Chemicals are chemicals, no matter where they come from. But because these “natural” substances have been consumed for decades, sometimes millennia, there is lower risk of safety issues. And the efficacy is predicted by the Machine. The risks are small and the potential benefits are substantial.
Read more about our in-silico paradigm in our How document (although we need to update it badly). In addition to providing a general recommendation to all people with Parkinson’s, this paradigm will facilitate the rapid prediction of custom combinatorial therapies and protocols for people according to their genetic makeup, stage of their progression (including for prevention), and symptoms. Personalized medicine will be revolutionized by this approach.
Initial efforts with a local, home-built machine
We started our work with PD by building our own machine, highly specialized for receiving, processing and storing data in the way that we needed. But at the same time being portable enough to transport for demonstration purposes, and meetings. It was small enough to fit comfortably in carry on airline luggage! We combined, designed and tested algorithms on that machine. We are still using that machine today, but it is used more as a controller, accessing all of the components in local hardware and in the cloud. We still call our entire system the "Machine", even though it has many components.
Just in case anyone is worried that our Machine will steal your identity or take over the world. You can relax. When we talk about "smart" or "smart computers" or "intelligent" or artificial intelligence (AI), we are not talking about human-like intelligence, with human emotions or drives. We are talking about things like machine learning and natural language processing. Highly specific problem-solving skills, yes. But evil, no.
Funding came just in time. Our growth.
The funding that Annette found for us enabled us to expand and grow the capabilities of our system. We were able to:
- Purchase larger, faster machines.
- Rent storage and processing resources in the cloud.
- Get access to much more data and systems information (including PD network analyses).
- Build live cell and live organism microarrays for training and validation.
- Dr. Ben and I had built a robotic array for testing and automatically "evolving" multi-drug therapies in the past. At the time we used components from an early 3-D printer. We are building a more advanced system for this project in order to teach the machine and to validate some assumptions. For other such examples of live traps and arrays, see:
- Modulation of alpha-synuclein toxicity in yeast using a novel microfluidic-based gradient generator.
- A programmable microvalve-based microfluidic array for characterization of neurotoxin-induced responses of individual C. elegans.
- High-throughput imaging of neuronal activity in Caenorhabditis elegans
- High-throughput, on-chip, whole-animal screening at subcellular resolution
- Outsource some of the engineering and testing, including algorithm and system design, and robotic array design and construction.
If you were one of the donors to this project then we thank you! Your contribution has really made all of this work possible.
We’ve accomplished so much in this past year, especially considering that we are all doing this in our spare time. I have a consulting business that keeps me busy. Dr. Ben is a postdoc, doing completely separate research. And he is not entirely comfortable with his boss and the other researchers finding out what he does in his spare time. Both of us were basically just tinkering with cool ideas when Annette came along.
Understandably, the paradigm of making predictions without testing, and making recommendations to sufferers, is controversial. We expect to ruffle some feathers among those who prefer established ways. Because of these realities, Dr. Ben and I would like to try to maintain a certain amount of anonymity and privacy for as long as possible. We hope everyone can respect that.
Introducing the World’s Best Parkinson’s Monitoring Station
One of the early achievements of this project was the creation of a Parkinson’s monitoring station. Basically, any data we can access that has anything to do with Parkinson’s and other neurodegenerative diseases passes through our parsing and extraction system. We don’t just access the standard data sets, we extract relevant information from text (see University of Delaware’s eGift project as an example) and even speech. Data sources include: research papers, patents (including applications), slide shows, presentations, posters, theses, datasets, analyses, videos, podcasts, comments, interviews, blog posts, Facebook posts, tweets, forum posts, etc. etc.
Information is pouring in every second of every day and is being processed by the Machine. It is, almost certainly, the largest monitoring station for Parkinson’s disease, and it is very possibly the largest of its kind dedicated to any disease, anywhere. The sophistication comes in our ability to extract useful information from all the data, the vast majority of which is absolutely useless.
Data, Information, Knowledge and Decisions
We are teaching the Machine about Systems Biology, especially pathway and network analysis, so that it can model PD computationally, as information is assimilated. For example:
- Signaling pathway cloud regulation for in silico screening and ranking of the potential geroprotective drugs
- Prediction of kinase inhibitor response using activity profiling, in vitro screening, and elastic net regression
- Pathways to neurodegeneration: mechanistic insights from GWAS in Alzheimer’s disease, Parkinson’s disease, and related disorders
Many research teams are working on various aspects of the network of PD pathways. The Machine is integrating those analyses, piecing them together to understand the whole picture. It then makes predictions on the sets of 4 or 5 pathways that likely need to be modulated in order to slow or stop PD progression. It then looks at all the evidence about OTC "natural" substances that modulate one or more of those pathways. See, for example:
- Towards a bioinformatics analysis of anti-Alzheimer’s herbal medicines from a target network perspective
- Network-based drug discovery by integrating systems biology and computational technologies.
- A Systems Biology Approach to Uncovering Pharmacological Synergy in Herbal Medicines with Applications to Cardiovascular Disease
So we end up with a matrix of options: multiple sets of pathways, and multiple combinations of substances for each set of pathways. We are working on a heuristic to judge each option in terms of likelihood of effectiveness, number of items, costs, difficulty and complexity, etc. The heuristic ranks the options for us and tries to explain the Machine’s decisions.
Wait for it…
Our goal is not to become a drug company. We are scientists, not businessmen. We have never sold products and have no interest in selling products. However, we are going to have to talk about natural substances and "supplements" that other people sell. These are OTC substances and somebody sells them. We’re not interested in a political debate about capitalism or the free market. Everybody sells. Drug companies sell, and supplement makers sell. Are these companies taking advantage of sufferers? Not if sufferers pick the right things that help them. Sellers simply provide options; we buyers ultimately make the choices. So let’s make the best choices possible. One of the main benefits of this project is that it takes much of the guesswork and uncertainty out of these choices.
So first we are going to take the Machine’s decisions, predictions, and figure out which practical options best match the predictions. Sometimes those predictions will require small lifestyle changes, like "get more sun" or "eat less sugar". Sometimes those predictions will require purchasing supplements. Dr. Ben and I will do our best to find the best real-world options. Annette will help us decide what recommendations are most reasonable in terms of convenience and costs. After all, her husband will be doing whatever we recommend for you.
What can you expect from us? If you are on our list you will receive our recommendations derived from the Machine’s predictions. We decided to release these recommendations in versions, just like software. In the next few weeks we hope to release version 0.0.1 (or whatever), which will then be followed with subsequent higher versions as new data and better algorithms improve the predictions. Also, Dr. Ben and I might change our minds about the best real-world substance or lifestyle change based on a prediction. Just because the predictions change doesn’t mean the old prediction was bad. It just means that the predictions tend to improve.
The first recommendation to be sent out will probably be only a list of two things. This is based on Annette’s recommendation to "keep things simple". It’s true that we think at least 4 or 5 pathways must be modulated to have any hope of stopping PD progression. That’s why many of the substances or lifestyle changes we will recommend are "multi-target". in other words, sometimes single items modulate multiple pathways. For example, see:
- Promise of Neurorestoration and Mitochondrial Biogenesis in Parkinson’s Disease with Multi Target Drugs: An Alternative to Stem Cell Therapy
- From Single Target to Multitarget/Network Therapeutics in Alzheimer’s Therapy
- Multi-Target Directed Drugs: A Modern Approach for Design of New Drugs for the treatment of Alzheimer’s Disease.
The Machine has identified a number of multi-target possibilities so far. In future versions, the number of substances or lifestyle changes will likely grow in order to maximize the likelihood of benefit.
That’s all for now. Please let us know if you have any comments or requests. And please share this post on all of your social media, in groups, in forums, on blogs, by email, etc. Thanks.
Researcher at Stop Parkinson's
Dr. Steve is a biochemist, specializing in medical bioinformatics and nutrition. Dr. Steve directs a biomedical consulting laboratory, focusing primarily on biomedical investing and health policy.
Latest posts by Dr. Steve (see all)
- Can Parkinson’s be treated using engineered viruses? - May 18, 2016
- Our Recommendation for Young Onset Parkinson’s Disease, Lite Version - September 23, 2014
- Our Recommendation for Acid Reflux (GERD) - September 16, 2014