BayMax: A Personal AI Healthcare Companion

Diagnosing diseases automatically has been an immense challenge, owing to their variable properties and symptoms. 

BayMax is a low-cost device for straightforward analysis and treatment of human diseases. By utilizing neural networks, the device can detect diseases and conditions, automatically, using end-to-end deep learning. It does so with an extremely high accuracy rate, relative to trained doctors. It can detect 1557 medical conditions, utilizing cardiac, dermatological, vocal, and symptomatic analyses.

This device can augment doctors by speeding up the time needed for diagnosis by pre-analyzing the user and providing estimated conditions. This scalable method of detecting anomalies before they pose a threat, holds the ability to create clinical impact around the world by profoundly increasing access and scope of medical care. 

Overall, this device will help alert physicians to high-risk patients, while making the doctors' analysis much more efficient and accurate; therefore, saving people, while decreasing costs and time.

More than half of all deaths in low-income countries are caused by communicable diseases, maternal causes, and nutritional deficiencies, called Group I conditions. 

More than 400 million people do not have access to essential health services, with an average of one doctor for every 2000 people in developing countries. 

Many of these people struggle to get enough food and thrive, so many do not have the time nor money to go to a distant hospital. In regions of Africa and India, there is a single doctor for every 20,000 people. Even more concerning is that these doctors usually centralize in urban areas, leaving rural villages with little or no medical support. Many within range of medical facilities still are not provided attention, as even the care is unable to keep up with the hundreds of people that require aid, resulting in further death. In low-income populations, the morbidity of diseases increases, due to overcrowding and increased transmission, and outbreaks of disease are more frequent and more severe when the population density is high. 50,000 people are dying every day from infectious diseases; in actuality, these diseases are easily treatable, but, unfortunately, many people do not have the treatment knowledge. 

From the research, three main types of users were identified:

1. Primary healthcare workers in underdeveloped areas of the world
2. Health care facilities/clinics using the device for telemedicine 
3. Home users concerned with their health

The healthcare workers in underdeveloped areas will use the device to get the preliminary diagnosis and prioritize the care of at-risk individuals. Even in developed areas, like the US, there is a shortage of physicians in certain areas (for example, in Pennsylvania's Coal Region). Using the device, local healthcare workers can identify the disease, conducting a wide array of analyses, to send to remote physicians for further approval. 

Home users, especially senior citizens, are reluctant to visit physicians due to limited mobility or unavailability of convenient transportation. Using the device they can conduct preliminary diagnosis themselves and the device can alert them and their caretakers of any serious health issues identified before problems arise.

Currently, BayMax provides four automatic, easy diagnostic analyses: symptom, heart, skin, and cognitive analyses. Each test takes around 30 seconds for the user, while the device takes around 5 minutes to process all the tests.

1. For symptom analysis, the user enters the symptoms and the neural network responds back with possible disease. This algorithm has an accuracy rate better than online symptom analyzers, such as WebMD. 

2. For heart analysis, the device has ECG electrodes which need to be placed on the hand and wrist. Just a few seconds of data is required to conduct an accurate analysis of cardiac conditions. 

3. For skin analysis, users simply point the device's camera at part of the affected skin, and take a picture. 

4. For cognitive analysis, all patients do is record audio of a particular series of sentences, and the algorithm can identify biomarkers of cognitive decline from speech.

The main goal of the tool is saving lives. By providing medical analysis to those who do not have access to healthcare, detecting diseases before the user even notices them through a 5-minute and comprehensive test, many people around the world could receive medical attention before their situation worsens. 

It will allow the population in less-fortunate areas more hands-on control over their medical data, as they will be able to independently conduct analysis. They will be able to identify any anomalies and proactively seek out medical care. Medical organizations will also be able to better identify at-risk individuals, as the device is capable of conducting patient triage to make sure that the people that really need assistance receive it.

Overall, BayMax has the potential to help many people around the world by democratizing access to medical analysis and allowing for a more efficient healthcare system.

The neural networks that have developed for this device are far above the current standard for automatic medical analysis. For cardiac and cognitive analysis, the algorithms are better than trained doctors, while, in general, the results of the tests are comparable to the rates of trained personnel. The algorithms also utilize a variety of neural network understandability measures, which allows the device to identify anomalous sections of patient data, not simply identify end disease, which makes the job of doctors much easier, by identifying anomalous sections automatically.

On top of this, the device costs less than $150, and due to the open-source hardware utilized in its construction, anyone, anywhere, can download the code for the algorithms and UI, and easily construct a version of the device in just under an hour. The algorithms have already been trained, so now they can be instantaneously deployed. This is much better than other implementations that require custom hardware, which delays production, and also are expensive. BayMax solves these problems.

BayMax is also entirely offline, which maintains the privacy of users. This is not a feature that many other implementations have, due to the difficulty in optimizing algorithms for offline analysis. 

Overall, BayMax prioritizes users above all else, making sure that people have access to the tool, while providing superior accuracy and comprehensive medical analysis.

The Deep NN (DNN) algorithm can identify 1557 various diseases, along with providing treatment advice. Biometric values such as oxygen saturation and electrocardiogram (ECG) values are calculated using a Recurrent NN (RNN), developed to detect anomalies: myocardial arrhythmias and ischemias. A Convolutional NN is on the device to identify and segment dermatological lesions. Vocal tone analysis, through an RNN, detects cognitive decline. These algorithms run on a Raspberry Pi processor. Once the four-step analysis has been completed for a patient (cardiac, dermatological, cognitive, and symptomatic), it conducts patient triage and generates a grouping of high-risk patients, so doctors know which patients to prioritize.

There will be three primary distributed channels: 

1. Users in underdeveloped areas - local manufacturing/local diseases
2. Telemedicine for developed areas with physician shortage - partnerships with regional hospitals and insurance companies, as well as home health care agencies
3. Home users - online sale and local pharmacies

BayMax would be accessible at local medical facilities, and, also by partnering with various humanitarian organizations, BayMax could possibly be spread further. By discussing with insurance companies, the health benefits of identifying diseases more efficiently is appealing for them, so in the future, we could partner with them to scale. 

For individual users, BayMax might possibly be able to be linked with their personal healthcare system, working with the doctors and hospitals directly to better increase their quality of care.

In the next year, we will make the device more compact and even more accurate, as technology evolves. The device will be connected with additional sensors to capture different biomarkers, including digital ones. The device will be widely available from online stores and also manufactured/assembled locally, and will also run on solar power to address off-grid use in underdeveloped areas. 

Over the next five years, the team will address three types of users: users in underdeveloped areas (local manufacturing/local diseases), telemedicine for developed areas with physician shortage (partnerships with regional hospitals and insurance companies, as well as home health care agencies), and home users (online sale and local pharmacies). We hope to grow the team to local communities to further build connections between medical organizations and populations in less-developed areas. 

The dream is that inside of every home, there will be a BayMax device to conduct daily analysis to identify any medical anomalies before the user even notices anything.

We are interested in scaling BayMax, and implementing the device. It is difficult to find resources to expand the tool and actually create the change that we want to make.

We will need have access to the resources to not only determine the process to begin implementation of BayMax, but also identify how to put BayMax into production to help even more people. Although the research into the technology behind BayMax has been worked on extensively, we would appreciate if we could have professionals in the field provide feedback, so that it can become an even better tool. Having mentors who have worked towards implementing these type of solutions would be an incredible asset, especially because, personally, we do not fully understand the pipeline to bring a technological solution like BayMax to the market. While the structure of BayMax's implementation has been identified, we would like to work with ToolFoundry to establish a business model to make sure that BayMax can continue to democratize access to medical care across the world.  The connections that Tool Foundry would be able to facilitate would also be an asset, as they would be able to link to professionals and experts to work with to further the progress of BayMax, not just technically, but also in its implementation.

We are interested in scaling BayMax, and implementing the device. It is difficult to find resources to expand the tool and actually create the change that we want to make.

We will need have access to the resources to not only determine the process to begin implementation of BayMax, but also identify how to put BayMax into production to help even more people. Although the research into the technology behind BayMax has been worked on extensively, we would appreciate if we could have professionals in the field provide feedback, so that it can become an even better tool. Having mentors who have worked towards implementing these type of solutions would be an incredible asset, especially because, personally, we do not fully understand the pipeline to bring a technological solution like BayMax to the market. While the structure of BayMax's implementation has been identified, we would like to work with ToolFoundry to establish a business model to make sure that BayMax can continue to democratize access to medical care across the world.  The connections that Tool Foundry would be able to facilitate would also be an asset, as they would be able to link to professionals and experts to work with to further the progress of BayMax, not just technically, but also in its implementation. 

3

We have experience in machine learning and data science.

Partnered with Akola University for preliminary testing.

There will be three primary distributed channels: 

1. Users in underdeveloped areas - local manufacturing/local diseases
2. Telemedicine for developed areas with physician shortage - partnerships with regional hospitals and insurance companies, as well as home health care agencies
3. Home users - online sale and local pharmacies

BayMax would be accessible at local medical facilities, and, also by partnering with various humanitarian organizations, BayMax could possibly be spread further. By discussing with insurance companies, the health benefits of identifying diseases more efficiently is appealing for them, so in the future, we could partner with them to scale. 

For individual users, BayMax might possibly be able to be linked with their personal healthcare system, working with the doctors and hospitals directly to better increase their quality of care.

We will initially focus on grants, and focus on partnerships with medical organizations to further scale the device.

We are interested in applying to the accelerator to scale BayMax, and for partnerships in the implementation of the device. It is difficult to find resources to expand the tool and actually create the change that we want to make.

Partnering with an organization such as Last Mile Health, or a similar philanthropic organization that distributes technological solutions to less fortunate populations, would be a great opportunity to further spread the BayMax device. We would provide the devices to them at minimal costs for implementation in one per village/area. 

We will utilize this prize to develop better AI models for automatic medical analysis, and also to test a few of the models in the field, in actual use-cases with doctors.

We will utilize this prize to develop better AI models for automatic medical analysis, and also to test a few of the models in the field, in actual use-cases with doctors.

We will utilize this prize to develop better AI models for automatic medical analysis, and also to test a few of the models in the field, in actual use-cases with doctors.