Early Intervention Systems

We have devised a software system that can predict when a patient/resident will become agitated and improve the overall experience for patients, by using a wristband, such as an Apple Watch. When a patient becomes agitated, he or she poses a risk to medical staff, other patients, and themselves. This costs US hospital systems, alone, close to $2.7 billion annually. By using biometric information gathered from the wristband and a deep learning algorithm, we predict when a patient/resident will become agitated. This wristband is combined with our companion application to store and visualize patient data safely on a cloud based system. Our solution extends to Covid-19 in elder care facilities such as memory-care and assisted living. People diagnosed with Alzheimer’s and Dementia are at most risk for Covid-19. We believe it will finally provide healthcare workers with a proactive solution to the systemic problem of patient agitation. 

The issue of patient agitation/violence has plagued the healthcare industry for decades and costs hospital systems, alone, close to $2.7 billion annually. These concerns include financial burdens, healthcare provider burnout/turnover, and diminished healthcare outcomes. While 75% of nearly 25,000 workplace assaults occur annually in healthcare settings, only 30% of nurses and 26% of emergency department physicians have reported incidents of violence in the hospital setting. This is due to an ineffective reporting system relying on passive observation and reactive measures. During the Covid-19 outbreak, Alzheimer’s and Dementia patients with the virus have shown to display  few of the common early warning signs of the virus and have only exhibited elevated levels of agitation. This means that agitated patients can often be overlooked even though they pose a greater risk of spreading the virus to other patients and staff. In fact, up to 69% of Covid-19 ICU patients exhibited agitated behaviors. Without Covid-19 the most agitated Alzheimer’s and Dementia patients can cost upto $30000 more to take care of. Thus, Covid-19 will only increase these estimates in the years to come and create an even greater burden for healthcare sites.

Our systems of predicting early stage agitation can improve testing, contact tracing, and proactive preparedness in these facilities. We are currently working with multiple assisted living communities to install a solution to identify early stage agitation. By identifying early stage agitation, we can improve the overall experience for residents while also improving safety and working conditions for nurses and caregivers. We are initially focusing on facilities specializing in Dementia and Alzheimer’s care as over 50% of the patient population display aggression and agitation. We want our proposed solution to increase proactiveness in preparation for a Covid-19 outbreak while building a scalable solution before a second wave. As a result, we have a timeline of 6 months. The wristband aspect of our product will provide an innocuous way to monitor high-risk patients that pose a danger to staff, other patients, and themselves. Lastly, the sensors will be integrated with software that will allow for live monitoring of patients while also providing frequent/relevant predictions regarding the likelihood of a violent outbreak. The motivation and theory behind our product are based on a large set of literature looking into means of detecting agitation using multi-modal sensors.

Our target industry is elder care facilities such as memory-care and assisted living. Currently in the US, 35% of all Covid-19 deaths are related to assisted living and nursing home residents and workers. People diagnosed with Alzheimer’s and Dementia are at most risk for Covid-19. However, they are not displaying the same symptoms as regular Covid-19 patients. Rather, many Alzheimer’s and Dementia patients are initially displaying Agitation as a symptom of Covid-19.  Our systems of predicting early stage agitation can improve testing, contact tracing, and proactive preparedness in these facilities. We are currently working with multiple assisted living communities to install a solution to identify early stage agitation to improve the overall experience for residents while also improving safety and working conditions for nurses and caregivers. We are initially focusing on facilities specializing in Dementia and Alzheimer’s care as over 50% of the patient population display aggression and agitation. We want our proposed solution to increase proactiveness in preparation for a Covid-19 outbreak while building a scalable solution before a second wave. We hope to share in ensuring the safety and health of all residents, caregivers, and staff, while providing family members with comprehensive information on their loved ones’ well-being.

Our service, which includes a wristband, deep learning algorithm, and companion software, provides a proactive approach that can facilitate public health preparation. There have been continuing studies showing that initial observable symptoms of Covid-19 in Alzheimer’s patients are only agitation. If there is a correlation between increased agitation and Covid-19 positive levels, our prediction models can lead to better testing and outbreak assessment at an earlier stage to isolate patients. The integration of our healthcare tools will lead to quicker and more accurate detection, while providing patient monitoring systems for timelier responses and improved care in these facilities. 

Our product is novel since it uses biometric information gathered from the wristband and a deep learning algorithm to predict a patient's agitation. This wristband is combined with our cloud-based application to store and visualize patient data safely. Our product provides healthcare workers with the first proactive solution to the systemic problem of patient violence. There are currently no direct competitors who create a commercially available solution to predict early stage agitation. Our solution would be able to change the lives of patients and healthcare staff.  Current reactive solutions include using sitters to observe potentially violent patients and to notify healthcare providers when a violent outbreak is imminent. These solutions represent the gold standard for ameliorating this issue. Nevertheless, these solutions will never be able to truly prevent violent outbreaks as they are not proactive. Our product is innovative for Covid-19 since we are able to predict a symptom, which is agitation, in Alzheimer’s and Dementia patients. This will lead to quicker testing and better preparedness for any outbreak in these facilities. With a vulnerable elderly population, every hour counts once there is a case of Covid-19 in long-term care facilities. 

Our product identifies early stages of agitation for a patient using a deep learning algorithm and smart wristband such as an Apple Watch. The wristband aspect of our product provides an innocuous way to monitor high-risk patients that pose a danger to staff, other patients, and themselves. The wristband allows for live monitoring of patients and predicting the likelihood of a violent outbreak. Since one of the primary focuses will be improving and scaling the prediction models. The various biomarkers that will be collected such as movement, hygiene, sleep quality, heart rate, and diet. The architecture will use de-identified patient information to ensure our system is HIPAA compliant. Once the algorithm/ model predicts a specific threshold value, dependent on the weights of each biomarker, it will alert the nursing staff about an imminent agitated event. The alert will be depicted through the notification service.. Preventative action in the figure below represents the pharmacological/non-pharmacological interventions nurses usually take to calm down an agitated patient. We are not a wearable company or an EHR company, but an AI and Analytics company that is looking to scale predictive algorithms to improve wellbeing and healthcare safety.

Deep learning algorithms have been used to predict specific clinical phenomena, such as episodes of delirium and palliative care interventions. Machine learning has even been used for patient violence risk assessment. Various machine learning algorithms have been used to classify and identify agitation using accelerometer data. Yet, these studies have only used motion capture data to classify agitation. Few studies have looked carefully into the pathophysiology of agitation, which could be due to its various and complex manifestations. Some universities, such as Yale University, are studying similar biomarkers on the detection of agitation such as heart rate variability, actigraphy, electrodermal activity and EEG in patients with schizophrenia. All of the links to studies we have analyzed are below and a demo video is also below. 

https://docs.google.com/spreadsheets/d/1rHB_Zs36cAJt_s1miUv7DF5AOc4GsJe0FhodYGi-KMo/edit?usp=sharing

https://youtu.be/V5eS2SALGLw

We spoke to nurses, caregivers, physicians, patients and families before creating our product. From these conversations, we identified barriers to entry, outcomes, and change. As a result, we focused on creating a solution to a problem that has long been neglected by healthcare executives and will directly impact everyday interactions patients have with providers. To ensure we achieve our longer term outcomes, making healthcare facilities safer and reducing violent outbreaks, we plan to continually listen to patient advocates and end-users. Each patient’s agitation profile is unique and requires us to carefully listen to the insights care providers have regarding agitation in various facilities. We are specifically focusing on a memory care facility so we can better understand agitation as it presents with individuals who share preexisting conditions. These memory care facilities have a very low barrier of entry for most technology since they are not encumbered by regulations often found in hospital systems. The site we are looking to pilot with is particularly unique because they have not implemented any technological solution to their facility and will likely depend on our product to provide automated patient well-being information without pen and paper. This means we have tailored our product to this pilot site in order to ensure greater retention amongst the care-givers and nurses and better collaboration moving forward. As we create a strong agitation prediction model, we will continue listening on the front lines of these facilities. The insights we derive from this pilot will help us  expand our model to the facilities within their memory care company. As we generalize our prediction model, we can transition to other healthcare facilities and ultimately reach our goal of providing proactive steps to intervene before a violent event occurs. We believe we can affect change by empowering and listening to providers every step of the way by creating a product that makes their own and their patient’s safety a priority.

Within the year we will be serving four memory-care dedicated facilities with 64 residents. Within the next five years, we will be installed in hundreds of these similarly sized facilities, as well as looking to license our AI and analytics into thousands of other similarly sized facilities. Furthermore, we hope to serve more individuals by eventually transitioning our service from an Apple Watch to environmental sensors. 

Within the next year, we hope to have a substantial impact on the detection of Covid-19 in elder care facilities. Our impact for the Covid-19 outbreak would be to predict early stage agitation in Alzheimer's and Dementia patients to improve testing and results in these facilities. Testing has been a large problem in this pandemic and we can hopefully create an AI tool related to this specific symptom before a second wave approaches. We hope to improve detection of a Covid-19 early enough in these facilities to ensure that the facilities can take proactive preventive measures early enough so that less patients and staff are at risk of getting infected. Our service can be easily transitioned to our long-term impact which would be to reduce patient agitation in all types of healthcare facilities to improve outcomes and experience. 

Within the next five years is to be the leader in detecting Early Stage Agitation and ensuring Well-Being to avoid ugly events or outcomes while providing a safer, healthier environment for residents and caregivers. We hope to facilitate patient monitoring through our AI tool development that can predict early stage agitation to improve the quality of care for residents/patients. Leveraging the increased demands of medical staff, such as ameliorating the pressures from the nursing shortage, is also one of our top priorities. With more efficient tools that enhance workflow and improve detection, our ultimate goal remains ensuring the highest quality of life for individuals with Alzheimer’s and Dementia. 

Acceptance is our primary concern. We have seen notification systems in the past that have been more of a disturbance rather than a benefit. Furthermore, our service architecture and experimentation are computationally expensive and require HIPAA compliant data storage. To overcome this challenge, we are using Azure cloud services and virtual machines to compute and encrypt the data we collect. Although our model network’s are computationally expensive, we have existing cloud credits to support our development and research efforts. In addition, we must prove that GRU-based models are more efficient than LSTM-based models in our analysis. Furthermore, teaching the nurses how to use the software in an efficient way that ensures comprehensive understanding may prove difficult.

We firmly believe our team’s collective expertise will help us overcome any unforeseen challenges. In terms of acceptance, we have identified that our service mainly focuses on finding a proactive approach, through our technology, to address and placate agitated behavior. Long-term competitive advantage will also ensure that our service continues to evolve with the changing needs of these communities, whether it be licensing, making changes to the software, or staying up to date with compliance. We have already established meaningful partnerships with our clients that have helped to improve the features of our software, such as helping caregivers attend to residents who may need attention and reduce any concern for notification fatigue. We will also develop long-term partnerships with associated service providers, such as insurance companies. 

Our team experience in machine learning and clinical design will also ensure that we are following appropriate IRB procedures and data science challenges in the future. While implementing our software may prove difficult, Karen Guccione, RN, BSN, has extensive experience working with and teaching nurses. Mrs. Guccione is the Director of Nursing, CEO, and Co-Owner of Safe Harbor Homecare, LLC. As a result, we will leverage her expertise to help with this alpha stage rollout. We are also going to help the nurses at the facility use our solution by providing educational modules that apply Nathalya Ramirez's pedagogical experience.

Our team consists of three full-time co-founders, one part-time employee, and six advisors.

Krishan Shah, recent GWU graduate in finance, will oversee our finances during our operation to help our team create a financial strategy for Early Intervention Systems. He worked at previous startups in the past, such as BookBuses, and as a Product Strategy and Innovation Intern at TradeStation. Nathalya Ramirez, now working as a teacher, has experience in nanofabrication and advanced statistical analysis and will help with prototype/algorithm development. Our CTO and Co-Founder, Rohan Patil, has experience in developing deep learning algorithms/predictive models.

Dr. Neil Sikka is an emergency medicine physician with over 12 years of experience in telemedicine and experience creating and selling his own startup. Dr. Chen Zeng, a Biophysics professor focusing on early prediction on the onset of diseases using computational simulations and deep learning techniques, will be overseeing algorithm development from the data collected at the facilities. Frank Hogan, who received his Small Business Development Certificate from the Wharton School of Business, will serve as a crucial part of our sales team. Mr. Nimesh Shah has 25+ years of experience in application development and data management for large corporations such as Dow Jones & Bank of America. Karen Guccione, RN, BSN, has over 20 years of experience as a Registered Nurse in many different healthcare settings and will assist in our clinical development and training. Jeff Borghoff, a fervent advocate of Alzheimer's Disease awareness, will serve as our patient advisor due to his diagnosis of Younger Onset/ Early-Stage Alzheimer's Disease at the age of 51.

Throughout our startup’s journey, we have grown as a team while committing to our mission and values. Once we won at the George Washington University New Venture Competition, we promised to give back to the community and gave us a shot at becoming successful social entrepreneurs. We volunteer at many of the GWU Entrepreneurial events and we won a $1,000 grant for GWU’s entrepreneurial programs in April of 2020. In addition, we  participated in the Alzheimer’s Association of Greater New Jersey Chapter's Mercer County Walk to End Alzheimer's on October 5th, along with Jeff Borghoff, an Alzheimer’s Advocate. EIS has created a team and has walked with our friends and family to raise $5,000 for Alzheimer's. In the end, we believe our ability to build communities within and outside of GWU has been our biggest accomplishment.

Our basic business model is a B to B sales on a subscription basis. We charge per patient per month for the wristbands and we charge monthly for the software on a SaaS basis. Our customer business model relies on monthly fees paid by its residents. We are initially targeting the memory-care facilities that have a population of 30-70 residents and already charge between $6000-$10000 per month for one resident. This model can be expanded to rehabilitation centers, assisted living, and home-healthcare. We also plan to add a licensing option for any facility that already has its own EHR/EMHR system in place. This option would allow us to enter into any medical facility in the future as our system will inevitably only use environmental sensors instead of wearable sensors. 

In providing the product and service, a team will present a demo and train staff on agitation, violence, and well-being. Once the installation starts for each location we will assign a clinical account manager who will facilitate communication between the caregivers and the technology team. They will also be in charge of resolving any immediate issues regarding issues with the system. As a result, each company will have their personal point of contact for any issues or concerns that may come up.

We are testing our solution at a Memory-care facility in 2020. We plan a test run in 4 facilities through 2020. Our goal would be to expand our services to more than 10 facilities in 2021. Most of our resources are paid through our start-up pitch competition winnings, which total to over $40,000 in prizes. We will continue to fund our project through non-dilutive funding as it will require significant time to further research and create a high true positive rate. If we do not secure enough non-dilutive funding from competitions and grants, plan a raising a seed round in early 2021. Our pricing model is $100 per patient per month without a wristband and $200 per patient per month with a wristband. Between the grant opportunities and revenue generated, we can create a solution that does not require a wearable and uses remote sensors/video data instead. This solution will be much cheaper and scalable to all types of medical facilities. 

MIT can help with most of the barrier I mentioned before in this application. However, there is one specific barrier that I want to address. Being originally from Colombia and a Women, it is very difficult for pitch competition judges and PhD's to take me seriously when pitching. There are already not that many women in AI and Technology. Some people think I am just another startup that made software or a medical device when in fact we are working with deep learning and analytics. For the most part, I am usually the only women in the meeting room at a tech meeting. The biggest barrier that MIT can help me with is helping Women Entrepreneurs in AI. There are some great advisors and faculty that can help me empower others and myself. On top of that we want to address the barriers above that are extremely important. The one barrier that SOLVE can help us with is improving the scalability of remote sensors. The prizes can also help us scale quicker while also improving contact tracing in Memory-Care facilities as soon as possible to help with the pandemic's ICU admissions. 

MIT has resources that many student entrepreneurs do not have access to. One example would be grant funding and grant proposal expertise. However, MIT's programs and staff go above and beyond that. Our main partners at MIT can help us scale and further develop our technology. It is imperative that his solution can be scaled efficiently as there is a clear for this tech in multiple verticals in healthcare. MIT as faculty that has extensive experience in scaling deep learning models that is extremely rare. Advisors that have this type of expertise would help us accelerate our current scaling models and improve outcomes in the healthcare industry. 

MIT provides many different types of programs for minority and tech entrepreneurs. The first organization would be the Martin Trust Center for MIT Entrepreneurship. Some of the advisors for that program such as Aarabi Balasubramanian, Mike Cavaretta, Aagya Mathur, and Jennifer Jordan, are just some extremely valuable people that can help us in our mission. All of these advisors have experience in helping women entrepreneurs succeed and have valuable insight in healthcare technology. The MIT Computer Science & Artificial Intelligence Lab has some extremely helpful faculty members that can help us with machine learning and healthcare analytics. There are too many people to list on this application, but John Guttag would be one of the Professors that could potentially help us in predicting early stage agitation. We also are interest in the MIT Enterprise Forum at Cambridge. This program focuses on using technology for social change. The program thinks of the small things that entrepreneurs need such as stipends and resources for grant writing which are very tough to find for student entrepreneurs. We would also like to work with any MIT Project/initiative focussing in patient monitoring, wellbeing, and Alzheimer’s technology if possible. 

My team and I have been working with memory-care facilities for a few years now. These are some of the most vulnerable populations for Covid-19. One particular facility we are working with was nearly brought to its knees with most the staff and residents infected. These residents had their life cut short and if our solution was fully implemented, we could have saved lives. I still stay up every night knowing that if I had more funding and resources available, I could have prevented people with Alzheimer’s and Dementia from dying alone in an ICU with Covid-19. This prize money will be used to scale our solution as quickly as possible to as many facilities as we can before and second wave comes. 

Our neuroscience research in agitation can help improves the quality of lives for residents, patients, and caregivers. This prize money will go right into research a way to predict early stage agitation in patients/residents without the use of a wearable. This step is imperative to ensure that we can scale to any type of medical facility that has an internet connection. By using remote sensors we can scale extremely fast with little cost outside of cloud subscription. 

As a women owned startup hat has devised a system that can predict when a patient will become agitated and improve the working conditions for nurses (90% are women) and caregivers (66% are women). The $75,000 prize would be used to scale and train medical staff to use our system in our 4 scheduled installations for this year and early next year. 

We are a wellbeing company, one of the UN's primary goals. Agitation does not improve the wellbeing for anyone in a medical setting. We can directly improve the lives of people who suffer from agitation episodes while also improving the safety of healthcare staff. The money from this price will be directly used in installing into memory-care facilities in 2021.  

We want to improve the quality of care patients and residents while also improving the working conditions for healthcare staff. Almost every human will walk through a medical facility and every healthcare staff are some of the most selfless and dedicated workers. By improving both groups we can change BILLIONS of lives all around the world,