Project BeeHive

What if we worked like a beehive?

Competition is the base of progress, that's why institutions are forever in an indefinite race to find solutions. Nevertheless, pandemics are no time for competition, and scientific authorities must function as one body, in simple words, like a Beehive! 

Nowadays, scientific research is highly dispersed and repeated, and the scientific community depends, by definition, on published research to build upon, which is a method of limited practicality, especially when timing is critical.

Project BeeHive is a smart database aimed to organize and direct our research efforts, as well as integrate it with AI technologies to find new drugs, predict patient outcomes, and anticipate viral mutations before they even occur.

Project BeeHive aims to tackle the following problems:

1.     The delay between study conduct, paper formatting, journal acceptance, and publication.

2.     The widely dispersed, or repeated research, especially during critical times (like pandemics).

3.     Wasted funds and human resources on “not so important” research.

4.     Buried ideas of bright but underfunded scientists, especially those from HINARI Group A and B countries.

How is that possible? BeeHive will:

1.     Map all research and Clinical trials about a specific problem into one circuit.

2.     Create a better “connection” between scientists, scientific institutions and funders, by providing a unified database for research targets and outcomes.

3.     Integrate Machine learning technologies to:

a.     Predict research paths of greater success potential.

b.     Predict outcomes in different populations (mortality, drug response, etc.).

c.     Predict upcoming Viral mutations and consequently: 1) Find existing drugs that fit these mutations OR 2) Design functional novel molecules.

d.     Provide an interface for the public to: 1) Participate in clinical trials near them OR 2) Donate their unused CPU power for large-scale computational analysis.

BeeHive is a Database with 4 Main functions, described below: 

Function 1: Research Collection and Mapping (Branching Logic)

Fig 1. The basic branching logic in research distribution and mapping

Fig 2. Example of a "SARS-CoV-2 Antiviral" research distribution Map

Function 2: Machine Learning (Realtime Prediction and Feedback loops)

- Deep Neural Network Technologies (eg. AtomNet) can be integrated with Beehive to map candidate drugs for experimentation, and to predict novel molecular structures.

- In addition, Decision tree statistical models enables us to:

1.     Predict research paths with greater potential, this provides a guide to scientists on “What’s hot!”

2.     Predict population outcomes (eg. susceptibility, mortality, drug response, etc.). Once again, feeding back into the main map to guide clinical trials.

Function 3: Machine learning (Inter-pandemic anticipation)

Beehive will utilize stored data to anticipate viral mutations, this gives us the chance to identify or invent drugs/vaccines beforehand.

Function 4: Involving the Public (Volunteering and Grid Networking)

Beehive involves the public by:

1.     Enabling them to volunteer to nearby research projects or clinical trials.

2.     Implementing Computational Grid Networking (eg. BOINC), where people can donate their unused CPU power for a good cause.

In simple words, Project Beehive has a main goal of shortening the timeframe from Case 0 to drug discovery. By doing so, it serves:

1. Humanity; The faster we find solutions, the better our chances of survival.

2. Underfunded Scientists; By giving them the chance to express their theories and ideas, which would otherwise require considerable financial and technical resources to execute.

3. Universities and Laboratories; By saving their financial and human resources from being spent on aimless research, and directing them to the experiments we actually need.

4. Funding bodies; By directing their funds to the most purposeful researches.

COVID19 isn't the first nor the last pandemic humanity faces. We need a unified Database that enables us to act synergistically during critical times. BeeHive could've affected our current situation in the following manner: 

1. More rapid and fruitful research outcomes, aided by the  organizational and predictive models integrated in Beehive.

2. A better implementation of funds, simply money well spent!

3. Surfacing of ideas from scientists who didn’t had a voice due to their lack of funds.

4. A model of predicted SARS-CoV-2 mutations, and a wide computational analysis in search for functional drugs.

Our ultimate answer to any pandemic is a scientific solution, be it a drug or a vaccine. The main defect that costs us human lives and considerable economical losses is our incongruity in research.

The ever expanding number of scientific journals and publications makes discovery even harder during critical times. Just consider the time it takes to format a paper, find an appropriate journal, send it to the journal, get peer reviewed, and depending on whether it gets accepted or rejected, you might have to resubmit it somewhere else! 

Now during this whole process, the world’s scientists are hurrying in different directions..

How many other labs might have already worked on the same idea? How much money is wasted due to this duplication? What if we actually needed to experiment something else to progress further? What about the scientists who can’t afford the thousands of dollars it costs to experiment or even publish their ideas?

To solve this defect from its root, we are in desperate need for a Pandemic Database that unifies us and puts our efforts in one direction. Once our data is collected and our research needs are mapped correctly, we can further integrate existing AI technologies to yield better outcomes.

I strongly believe that Beehive is our answer to every possible outbreak, be it the seasonal influenza or the disastrous pandemics, together we win.

Beehive itself is a database expressed through a website and an application (For Mobile or Computer devices). The Database Management System will implement a simple branching logic sequence (presented in the next section) to organize research maps. 

Once this database is in place, Decision tree learning technologies can work in realtime to predict research routes of better success potential.

Deep neural networking technologies are also integrated with BeeHive to predict viral mutations and match molecular candidates for in vitro trials.

Computational Grid Networking is another technology used to aid our computational processes, by enabling people to donate their unused CPU power. 

Evidence 1: The Database

The best evidence here is the Cochrane Library, a universal database that effectively organizes systematic reviews. 

Cochrane: About Us: https://www.cochrane.org/about-us

Evidence 2: Branching Logical Sequence

a.     Target A presents the basic sequence of target mapping and booking.

b.     Target B is completely free for “booking”.

c.     Target C is successfully experimented in vitro and is available for clinical trials “in vivo”.

d.     Target D is fully experimented, both in vitro and in vivo, results are available for users, who will have an option to add their suggestions/modifications.

e.     “Add+” is where scientists can propose their theories, if accepted via peer review, it can be mapped as a target.

Fig. Booking a research with our virtual lab

Evidence 3: Decision tree learning 

Divyashree S. , Divakar H.R. Prediction of Human Health using Decision Tree Technique. Int J Comp Sci Eng, 2018, 6(6), 805-808. https://doi...

Evidence 4: Deep Neural Networks

In Prediction of Viral Mutations:

Salama, M.A., Hassanien, A.E., Mostafa, A. The prediction of virus mutation using neural networks and rough set techniques. J Bioinform Sys Biology 2016, 10 (2016). https://doi.org/10.1186/s13637...

In Prediction of Drug structures: 

AtomNet uses convolutional neural networks in their software to design drugs, their technology is successfully used with Ebola; https://www.atomwise.com...

Evidence 5: Computational Grid Networking

BOINC is a technology that enables effective research via CPU power sharing, it has been used in many famous projects, like the Einstein@home and the IBM World Community Grid.

BOINC: https://boinc.berkeley.edu/index.php

IBM WCG: https://www.worldcommunitygrid.org/discover.action

BeeHive is currently in the Concept stage, therefore it still doesn't serve. Once functional, BeeHive serves different communities at different times, be it the whole world during a pandemic, a single country during an epidemic, or even a small community during an outbreak.

Within the next year: 

• Gather necessary funds 
• Create fruitful partnerships
• Publish the Beta version of Project BeeHive.

Within 5 years:

• Have a functioning database that is internationally accepted 
• Achieve the short and long term outcomes aimed for in the aforementioned theory of change

Two main barriers face Project BeeHive; Funding and Partnerships.

BeeHive essentially requires considerable funds for the main setup; including servers, hosts, database management systems, and hired personnel.

In addition, international acceptance is essential. BeeHive aims to establish partnerships with health authorities like the WHO and major publishers like Elsevier to power our data input, as well as tech partners like IBM and Microsoft to power our machine learning technologies. 

Furthermore, since BeeHive essentially eliminates the process of publication during pandemics, it is expected to face some resistance from authors who would like to see their work officially published, and of course, from the journals themselves.

1. Funding and international acceptance:

Project BeeHive is first revealed on MIT Solve, through which I hope to secure part of the funds and gain international acceptance. 

Of course, further efforts will be put in place to acquire further funds if needed and to maintain a relationship with international healthcare authorities. However, MIT Solve remains a golden ticket for our progression!

2. Authorships and Publication:

BeeHive will adopt a policy similar to Cochrane's, where the experiments on the database are considered published work with author's name's and institutional affiliations. Additionally, authors may also claim the right to rewrite their finding in a Journal-compatible manner and resubmit it for publication elsewhere, as long as their data is live on BeeHive. 

At the moment, none. Depending on raised funds, BeeHive may involve tens of programmers, data scientists and software developers.

For quite some time, BeeHive was no more than a plan that I sketched down the corners of my cortex, simply because I knew I needed experts, funders, and support from major institutions to bring it to life. When COVID-19 surfaced, it costed us a lot of innocent lives and economical burden, which made me feel helpless being unable to deliver a solution that might impact this crisis.

Solve is like a golden ticket to BeeHive, I get to meet the experts, gain funds, mentorship, and partner with strong institutions. I truly hope that soon enough I can witness the impact of BeeHive on our world. 

Setting up a big idea like BeeHive is not easy, no single person can achieve that on his or her own. Therefore, BeeHive must partner with institutions capable of driving our development and improvement. In addition, we definitely need experts in multiple fields, including database science and machine learning. 

BeeHive is open for partnerships with research bodies, tech parties, healthcare authorities, and all funders looking to create a substantial improvements in the way we tackle pandemics.

Health authorities: WHO and CDC

Tech institutions: Microsoft, Google, IBM ..etc. 

Academic institutions: MIT, Harvard, Oxford..etc.

Scientific Publishers: Elsevier, Springer, Wiley..etc.

Data science, artificial intelligence, and machine learning are the 3 main elements of BeeHive! These 3 elements conceptualize a database that aids us during crisis, be it a small-scale outbreak or a worldwide pandemic. The technologies to be used include, but are not limited to:

• Database Management Systems
• Supervised and Unsupervised Machine learning
• Tree Decision technologies
• Deep Neural Networks
• Computational Grid Networking

Apart from the previously mentioned goals of BeeHive, on the long term, I hope to further implement machine learning to create a "Screening Service" for developing countries during pandemics, where the healthcare personnel can simply upload X-rays, CTs or even Lab values to BeeHive to stratify if the patient is infected and at what stage is he in. Of course, we should take it one step at a time, since setting up the primary database itself is costly.

The AI for Humanity Prize will definitely be a huge leap for BeeHive. Not only will it enable the primary set up of the database and its associated technologies, but it will make BeeHive a presentable functioning solution, that can strongly present its services for partnership with international institutions and healthcare authorities.