The Pain and the Plan
Parkinson’s disease is the slow disintegration of everything you have built in your life. There are drugs to reduce some of the symptoms, but doctors have nothing that can stop, or even slow down the daily loss of brain cells. Advances towards a cure have been painfully slow because Parkinson’s disease is extremely complex. As of this writing, doctors do not know where in the body Parkinson’s starts, how it spreads, or even how to accurately diagnose it.
Because of the degenerative nature of Parkinson’s disease, the daily loss of brain cells will make the any future cure even less likely for today’s sufferers. Parkinson’s sufferers do not have the luxury of waiting a decade for a clinically “proven” drug. They need the best option available today!
Therefore, the mission of this project is to quickly find and publish a therapy that is most likely to slow brain cell death. Using a new treatment paradigm involving artificial intelligence, a library of existing, safe substances, and no safety or efficacy testing, we think this can be done in a matter of months rather than years. In this way, Parkinson’s patients will have the best possible chance to stay healthy and alive long enough for a cure to be found.
Parkinson’s is a devastating and debilitating disease characterized by massive death of dopaminergic brain cells leading to the visible shaking, stiffness, slowness, and instability. Even more disruptive are the non-motor symptoms largely unrelated to dopamine. These include sleep disturbances, constipation, bladder symptoms, sweating, sexual dysfunction, fatigue, pain (particularly in a limb), tingling, lightheadedness, sensory dysfunction (esp. loss of smell), mood disorders (including depression, anxiety and apathy), psychosis, cognitive impairment, dementia, and often a premature death. There is no cure. There are no treatments to slow the degeneration. There are only drugs and other therapies designed to reduce motor symptoms and comfort sufferers as their brain cells continue to die unabated.
The precise causes of Parkinson’s are not known, but there are a number of hypotheses and a number of proposed pathways that progress to the disease diagnosis. Environmental and genetic factors both appear to play a role. There are many fine research labs around the world trying to understand the precise mechanisms in Parkinson’s initiation and progression. Such detailed understanding will be necessary for finding a cure.
Physicians will often talk about treatments for Parkinson’s disease itself, but this is misleading. Doctors prescribe drugs only for reducing motor symptoms and relieving some of the other suffering of patients. As the disease progresses, and the severity of symptoms increase, greater quantities and varieties of drugs are required. More drugs are prescribed to reduce the side effects of previous drugs. The drug regimen becomes more complex as the degeneration progresses. For example, see this post about hockey coach Steve Ludzik who needs to take 22 pills to control his symptoms.
Science is a slow, methodical process, but there is little doubt that eventually all the precise causes of the disease will be determined, and a treatment found that prevents and cures all the Parkinson’s disease varieties. We all hope that this day comes as soon as possible, but it is very likely a decade or more away.
The obstacles to curing or slowing Parkinson’s are:
- Parkinson’s is a complex, degenerative disease that has been a huge challenge for researchers.
- Drug development and testing is a slow, expensive process often requiring one or more decades.
- Lack of good Parkinson’s disease model systems makes screening difficult.
- Parkinson’s complexity involves multiple pathways, likely requiring a combinatorial therapeutic approach.
- Drug combinations are extremely difficult to predict and test, as the options are endless.
- Multiple Parkinson’s subtypes may each require separate therapies.
Sufferers and their loved ones are faced with a conundrum. With the help of drugs or surgery they are attempting to manage motor symptoms. But while sufferers are taking their drugs, and hoping for a cure, their brain cells are dying unabated. Their chances of being cured depend on the stage of their disease, the speed of degeneration, and the date of a future cure. So it is only logical that Parkinson’s disease activists, researchers and physicians should focus on:
- Earlier detection,
- Slowing the degeneration,
- Finding a cure faster
This project is focused on slowing the degeneration because there is a huge gap in the options provided by physicians. Physicians tend to wait for the approval of their medical organizations before considering any therapy, but sufferers need the most promising options NOW, generated using all the scientific information available. Brain cells are dying at this very moment, so patients don’t care that the options are not perfect or approved.
Arbitrary and Desperate
Many Parkinson’s sufferers are not aware that their brains are degenerating at the same rate, with or without medication. They are either resigned to their fate or falsely believe that the medications their doctor prescribes is treating their disease. Others sense the gap in treatment and realize they have to take matters into their own hands if they have any hope of slowing the ongoing death of their brain cells.
Patients and their families are forced to search the internet for ideas. They often end up buying rather arbitrary nutritional supplements. Sometimes they read a story in the news, or a scientific article that mentions a supplement. They may see an advertisement in a magazine, or a discussion on a blog or forum. These haphazard selections of supplements speak to the desperation sufferers feel.
Many choose treatments like they choose spiritual cults, putting their faith behind a popular alternative treatment or charismatic personality. This desperation is understandable given the lack of real medical options, but we hope to take a bit of the guesswork out of these decisions.
Bridging the Gap
Parkinson’s disease has no cure. There is no treatment that even slows the disease, though sufferers are often not made aware of this by their doctors. Patients are given drugs that only reduce symptoms, while brain cells continue to die unabated. Sufferers and their loved ones need a solution for that huge gap between relieving symptoms on the one hand, and a future cure on the other. The only way to bridge the gap is to slow or stop the brain cell death so sufferers remain alive and healthy long enough for a cure to be found. They deserve the best available option based on all the scientific information available. Let’s find that option today and give our loved ones a fighting chance!
Despite the lack of a complete understanding of the causes of Parkinson’s disease, there are hundreds of research papers in the scientific literature describing potential treatments. These papers describe the use of dozens of chemical compounds in Parkinson’s animal models, cell culture and other neurodegenerative disease models. Many of the chemicals used in these studies involve natural compounds or plant extracts that are already considered GRAS (generally recognized as safe) by the American Food and Drug Administration (FDA), or are exempted from GRAS, or are existing food additives and dietary supplements.
What we need is a systematic, rational approach for predicting the value of common supplements and other available compounds for sufferers of Parkinson’s disease. We need to find the best combination of compounds, and daily quantities of each, in order to achieve the maximum benefit.
The difficulty in making these determinations is in finding useful patterns from the copious amounts of medical and biochemical research data, including sequence and structural datasets. If one has enough data, and the expertise and computing power to process it, one can spot patterns and links that no one else has noticed. From these patterns, rational predictions can be made about the best possible combinations and quantities of compounds to slow the brain cell death.
The technical name is: In Silico Combinatorial Therapeutic Prediction.
Here is the simplified version:
- These days, powerful computers are smarter than we are. So if Parkinson’s disease is too complex for humans to understand, ask a computer.
- Ask the computer to consider substances that already exist, and that have already been studied in neurodegenerative disease research.
- Further, ask the computer to consider only those substances that are safe and can be purchased over the counter. Most of these substances are considered “natural”.
- Ask the computer to predict the most likely combination of these substances for interacting synergistically with multiple, complimentary Parkinson’s pathways, and in multiple Parkinson’s subtypes.
- Because these substances are safe and natural, do not test them, do not study them. Simply offer the information immediately and freely to sufferers and their doctors.
- Modify the information frequently as new data is received and hardware and software improved.
This paradigm is described by the following principles:
- Incurable Disease
- Degenerative diseases are complex: cures are improbable within the lifetimes of living patients.
- At the very least, patients want to slow the degeneration and retain as much of their function as possible.
- Traditional drug discovery methods have not been successful at creating disease modifying treatments.
- Traditional methods that seek to synthesize new drugs, which must then be tested on a succession of likely inadequate disease models, followed by human trials, followed by an approval process, will take many years.
- In summary, degenerating patients do not have time for:
- Synthetic drug design
- Testing on escalating disease models
- Multiple rounds of human clinical testing
- The safety and efficacy approval process
- Therefore, we propose a process involving NO DESIGN, NO TESTING and NO APPROVALS
- Combinatorial Therapy
- Degenerative diseases are complex, multifactorial viscous cycles that likely require multiple, synergistic compounds interacting within multiple pathways, in a combinatorial therapy approach.
- The number of possible combinations of substances are almost infinite.
- One drug is difficult enough to test through the traditional process. There are only enough resources to test a few combinations per decade.
- Therefore, we propose the rational prediction of COMBINATORIAL THERAPIES
- Computational Prediction:
- Computers and algorithms that run on them have reached a level where they are routinely finding complex patterns in data that humans never could have found.
- Computers are predicting outcomes and probabilities in extremely complex systems, yielding often surprisingly useful results.
- It’s time to set computers on the task of predicting therapies for complex diseases like Parkinson’s disease.
- Provide as input to the computer all the existing scientific data we have on the disease, plus any other texts, discussions, videos, etc, by researchers on the subject.
- Therefore, we propose the use of ADVANCED ALGORITHMS towards the prediction of therapies for complex diseases.
- OTC Safe Substances:
- For each incurable disease there are hundreds of scientific articles written about the efficacy of many dozens of existing, over the counter (OTC) substances.
- These OTC substances are often phytochemicals or other “natural” substances, each with active chemical compounds having established medical relevance.
- These OTC substances are GRAS, generally regarded as safe, or otherwise not requiring additional safety testing.
- These OTC substances constitute a vast therapeutic chemical library that are both safe and medically relevant.
- Therefore, we propose the use of only EXISTING, SAFE SUBSTANCES that do not require research and development, safety testing or approvals.
- Action Steps:
- Develop. Build a computer to run the necessary algorithms to examine all existing, relevant scientific data, including unstructured data.
- Predict. Have the computer make predictions about the efficacy of combinations of OTC substances.
- Publish. Publish those predictions for optional use by patients and their doctors.
- Improve. As new data emerges, algorithms improved, computers made more powerful, constantly re-run the algorithms and publish a new version as predictions are modified.
- We propose the rapid and rational prediction, using advanced pattern-finding algorithms, of combinatorial therapies consisting entirely of existing, safe substances for the purpose of slowing progression of degenerative diseases that involves no safety or efficacy testing.
- Our plan is to use existing and novel algorithms to find hidden patterns within the scientific literature and databases that suggests a combination of existing, safe, natural compounds that would likely be most effective at slowing Parkinson’s disease progression.
- Taking cues from existing medical precedent, this combination will be in the form of multiple supplements and other compounds that attack the problem from multiple angles simultaneously. These “angles” correspond to disease pathways that, when present in combination, lead to Parkinson’s.
- Parkinson’s disease is highly variable, owing to the fact that there are multiple “points of failure” leading to a Parkinson’s diagnosis. A number of subtypes have been proposed. The algorithms must take into consideration this variability in order to benefit the most people.
- Based on the results, we plan to freely publish a set of recommendations designed to have the greatest likelihood of achieving these goals.
- The set of recommendations will be updated regularly, and distributed exclusively by email, as new data becomes available, and as we are able to improve the analyses.
- Scientific and medical research gets published in the form of text and images in specialized magazines. Raw data, for example DNA and protein sequences, is stored in databases and obtained as datasets. There is so much published information and raw data that the human brain cannot possibly find the useful patterns in it. This is where computers can be helpful.
- Computational analysis of biological information is called Bioinformatics. Bioinformatics uses fast algorithms and powerful machines to find useful and interesting patterns in biological data that the human brain is not powerful enough to see.
- Our scientific backgrounds consist of an unusual combination of skills, including bioinformatics, statistics, epidemiology, recommender algorithms and other information filtration algorithms, internet programming (including web crawlers), biochemistry and human nutrition (especially phytochemicals). This is the kind of cross-pollination that makes this particular innovation more likely. That said, any reader who feels they can contribute in any way is encouraged to get in contact.
- Algorithms can make predictions about humans better than humans can. More and more, algorithms are seeing patterns in scientific and cultural data that people cannot, in areas as diverse as crime fighting, to medical treatment strategies. Read more about such examples:
- A large-scale evaluation of computational protein function prediction
- L.A. Cops Embrace Crime-Predicting Algorithm
- Winning at “Jeopardy” using tunable confidence levels of evidence
- An algorithm for predicting the optimal treatment strategy for a disease in newborns.
- MIT prof and student discover algorithm for predicting trending Twitter topics
- A machine-learning algorithm that predicts the popularity of songs.
- We intend to apply similar techniques to make predictions and find these hidden links and patterns in Parkinson’s related data, including literature and discussion about “natural” substances that interact in related systems.
- More specifically, we intend to combine an evolving Parkinson’s “Connectivity Map“, with an hypothesis generating and weighting algorithm that simulates the scientific method itself, with ideas from Evidence-based Medicine.
- We further intend to use linguistic algorithms to parse the texts of research articles in order to look for connections not found in the datasets.
- We also feel that some of the most useful connections can be mined from the minds of researchers as they give news interviews and lectures, or write blog or forum posts. Often they feel more free to hypothesize in less formal media.
- Multiple algorithms are being adapted, combined and tested on subsets of the data, including text-based sources of information.
- We live in the era of big data, but that doesn’t mean that all the data is freely available. Part of the challenge is getting the data, collecting data on an ongoing basis (as more data is generated every minute), gaining access to the data, getting permissions, of finding other ways around limited access to data.
- Some of the data is in a format that can be obtained as datasets for feeding directly to the algorithms. However, much of the data must be obtained using web crawlers that explore the internet, looking for relevant content.
- With ongoing access to the data, the algorithms will parse the data, looking for factors that link hypotheses, models and compounds. In the vast datasets, tiny significant signals appear. And it is these signals which hold the clues to predicting the most promising compounds.
- Challenges: Since this is a private, non-academic effort, access to commercial data becomes an issue. Also, servers and bandwidth required to crawl large amounts of online content must be obtained. We are currently working to overcome these challenges.
- This work is being done outside any academic or corporate context. Even if it were permitted to do so, we want to limit any potential conflicts of interest. Therefore, all equipment, services and access to data are being obtained privately. All computation is being performed outside of business or academic facilities.
- While some of the calculations can be performed on servers in the cloud, some of it must take place on a local machine. The machine must also be portable enough to allow ease of travel.
- To this end we are publishing our progress in building and testing a small but powerful machine capable of storing and accessing the data, running the algorithms and crunching the data very quickly.
- The machine needs to:
- be very fast
- have a huge memory
- have massive data storage capacity
- be small and portable
- be robust, can take 24/7, hot operation, with constant air circulation.
- Raw results from the initial computations will not be in a directly usable form. Decisions will need to be made. Most of the technical decisions will be with the help of other algorithms, but some of the higher level decisions will be with the help of people like you. We will seek the input of community members for many of these decisions, so we have created a group on Facebook for these discussions. We will extend invitations to those with the enthusiasm and participation that catches our eye.
- The next step will be to further analyze results and extract relevant and usable information.
- Extract a list of compounds and lifestyle modifications suggested by the analysis.
- Cluster the list according to Parkinson’s pathways targeted.
- Consider Parkinson’s subtypes, etiologies, disease stages, and other hypotheses.
- Choose an initial subset of this list where each compound:
- Is available as a nutritional supplement, over the counter compound, or is available in certain foods or food products.
- Is designated GRAS (generally regarded as safe), or otherwise accepted for human consumption by the FDA
- Covers a wide range of pathways, hypotheses, stages and subtypes
- Is reasonably inexpensive
- We will make the final decision to publish the recommendations based on one single criterion: that we would, without hesitation, follow the recommendations ourselves and encourage our loved ones with Parkinson’s to follow the recommendations.
- Results will be released as a set of recommendations to the public. Recommendations will be in the form of three different options in order to maximize usability of the information.
- The multiple options will minimize any objections in terms of cost and intervention type. Each has its own benefit profile:
- A list of dietary and lifestyle modifications designed to achieve at least 10% of maximum benefit (see bottom). This will be the route of choice for those who cannot afford supplementation, or who reject the idea of supplementation. Limited to simple and relatively inexpensive lifestyle modification. Expenditure of $5 per day maximum.
- A list of existing supplementation sources and directions designed to achieve at least 40% of maximum benefit. $10 per day maximum.
- A novel formulation with simplified regimen designed to achieve at least 80% of maximum benefit at $10 per day maximum. We will then take this formula to a manufacturer. Format and regimen will be determined with the help of the advisory forum.
- Recommendations will be delivered by email, not by blog post. Sign up now to receive the recommendations as they emerge and are updated.
The Follow Up
- Sufferers and their loved ones are encouraged to provide survey information and feedback, including qualitative observations and, of course, any metrics that are available.
- We will create a survey and feedback system to more efficiently track user progress and any issues or suggestions users may have, and integrate those suggestions into the recommendations accordingly.
- We recognize that this will not constitute proper experimental data, but without reliable metrics and the budget to do the testing, this is the best data we can get at this point. And remember: our goal is speed of delivery using existing, safe substances. Sufferers will not wait for clinical results. Furthermore, the constantly updated recommendations and versioning system will make any testing of the latest recommendations impossible.
- We will post on the blog any news regarding progress on the development of reliable metrics for Parkinson’s disease.
- Feedback will help us spot any trends and improve the recommendations and methodologies.
- No recommendation is final. As new data are produced, algorithms improved and feedback received, the recommendations will be updated. We expect to use a versioning system, similar to software. You can view all results and updates in your email inbox.
- Maximum benefit is an evolving assumption of the maximum theoretical benefit with no constraints on expense, access to rare compounds, and inconvenience in administration. The number is primarily for our own use but it may be interesting to others. Daily expense maxima will be influenced by discussions on the advisory forum.
- Use of this information is conditional on your agreement with the Terms and Conditions.
Like this Project? Here are two quick but important things you can do:
- Sign up to recieve the results. The machine’s results, the recommendations and all updates will only be sent out by email. Please do it now.
- Get the word out. Please send an email to everyone you know who was diagnosed with Parkinson’s disease AND their loved ones (who sometimes have influence on health decisions). If you are on Facebook or Twitter or LinkedIn, connect with this project there. And share our pages with everyone who might be interested. Remind them to sign up to recieve the results. Please do these now.