A Piedi

PROBLEM: The global leading cause of disability in older adults (40% prevalence) is osteoarthritis, and the most effective non-surgical intervention is physical activity. Yet, a staggering 87% of OA adults are non-adhering.
This is because physical activity modalities are non-tailored and derived from subjective questionnaires. When the mobility recommendations are over or under-estimated, pain is exacerbated, discouraging the patient from adhering to it.
 
SOLUTION: Empowering OA patients with a gait-analysing mobile software that calculates the modality of physical activity per patient, and connects a network of other OA walkers for peer-to-peer walking recommendations at convenience to their whereabouts.

GLOBAL REACH: The prevalence of OA is increasing due to aging populations and a rise in obesity. By 2050 people aged over 60 will account for more than 20% of the world’s population: 130 million older people will suffer from OA worldwide, of whom 40 million will be severely disabled by the disease.

40% of older adults (+65 years) have symptomatic osteoarthritis (OA), the leading cause of disability in this population worldwide. OA is also a contributor to comorbid non-communicable disease emergence, such as diabetes and heart disease, which are the highest causes of death. Whereas no cure for OA exists, the best proven non-surgical mitigation and management method for OA is physical activity. The correct level of physical activity is proven to mitigate pain and prevent muscle atrophy, thereby maximizing joint function. Yet only 13% of OA adults meet their physical activity requirements.
This is because – today – orthopedic physicians rely on a subjective questionnaire to estimate an OA patient’s pain levels, and build the recommended degree of physical activity on this estimate. When the physical activity recommendations are over or under-estimated, pain is exacerbated. This leads to a number of other complications common the OA population: further limited mobility, lower extremity muscle atrophy, sleep disruptions and mood distress, and others. These initiate a downward spiral of mobility discouragement with OA patients. Furthermore, recommendations simply state what a person should do (e.g., number of minutes of walking) and not when/where/how to achieve those goals as part of daily life activities. 

• The patients
  The symptomatic osteoarthritic population is 21 million in the US alone. Worldwide, it is a population of 630 million. 
• The orthopedic physicians: The physicians of OA patients worldwide
• The carers: The health and family carers of OA patients worldwide in the clinical and home settings

A gait-analysing software installs on the patient's phones, and operates in the background. It calculates the modality of the osteoarthritis condition, and builds on it a recommendation of the correct amount of physical activity (i.e., walking) needed to diminish knee-pain and maximize joint function. The learning algorithm combines with a social network with other OA walkers, leading to peer-to-peer sensor-driven walking recommendations at convenience and proximity to their neighborhoods.

Our system also includes a world-first technology for the simultaneous collection, interpretation, application of human mobility and activity data - seamlessly sourced from smartphone sensors and processed using
advanced statistical and probabilistic models, data-aggregation methods, machine learning algorithms and AI techniques. Our technology creates highly precise and fine-grain information about people's daily travel routines and movement patterns - it enables us to understand when and where people travel, for what purpose, on what route(s) and using what mode (i.e., walk, bike, car, bus, tram, or train). 

Combining the knowledge of the desired levels of walking for OA patients with the detailed understanding of what each person does on a daily basis enables us to provide recommendations for physical activity that are commensurate with individual capabilities/limitations and embedded into each person's daily routines. Our technology will provide fully personalized recommendations and address the two main challenges for achieving safe/desirable levels of physical activity for people with OA - (1) inappropriate recommendations that exacerbate pain and result in patients progressively exercising less and less, and their health further deteriorating; and (2) requirement for people to set aside time exclusively for exercising (lack of time, for all demographic groups, has been consistently identified as a critical barrier to exercise).

We recognize the social factor in behavior augmentation. We are not only a smart software that builds walking recommendation on gait analysis; but we also include carers and loved ones in the conversation through SMS and tailor-made mobile applications specifically developed for older adults. The social engagement – often overlooked by technical solutions – is at the heart of our joint degeneration mitigation mechanism. 

As a way to increase physical activity, routine travel is appealing because it does not require setting aside time exclusively for exercising. Lack of time, for all demographic groups, has been consistently identified as a barrier to exercise. When the behavior is ingrained in daily activities, it requires a lower commitment threshold and behavior maintenance is achieved. To date, behavior change interventions to promote physical activity have yielded highly disappointing results (particularly for individuals with chronic diseases and conditions like osteoarthritis) as they require people to set aside time for exercise, provide generic recommendations and ignore people’s daily travel routines.

Apiedi provides fully personalized recommendations for physical activity that are commensurate with the capabilities and limitations of OA patients, and are embedded into their daily travel routine. This, together with embracing the social factors of behavior change, is what makes our technology innovative. And it works!

The Biomechanical Detection system that we are developing (with the Motion Analysis Laboratory of the Harvard University Medical School) uses mobile sensor data and self-adjusting algorithms (based on Random Forest classifiers) to analyze an individual’s walking cadence and gait patterns. Artificial Intelligence techniques are being used to determine fully personalized physical activity recommendations based on ‘pain thresholds’ captured through analysis of smartphone sensor data. More specifically, the system will automatically: 

1) detect the biomechanical condition of individuals (e.g., knee osteoarthritis) based on walking patterns

2) determine personalized, safe, realistic and desirable levels of physical activity (i.e., how much an individual can and should walk per individual event and on a daily basis) to achieve desired health goals. 

We also employ Artificial Intelligence in the Activity Detection system that we have developed (and that we will use to identify what, when, where, how and why people travel). Our system uses unsupervised learning techniques to detect and predict travel patterns in real-time. More specifically, our system identifies to a high-level of precision the true origins and destinations of all trips, the travel mode used (e.g., walk, bike, car or transit) and the true door-to-door travel time (a significant limitation of current systems that ignore, for example, the time spent looking for parking and the walking component at the beginning/end of car trips). Furthermore, our system includes adaptation of the DeepCity learning framework to map destinations (even in the presence of noisy location estimates) based on an individual’s past behavior. 

The reasons why OA patients do not adhere to physical activity programs are: 

1. Recommendations for increasing walking levels are generic and do not consider individual pain thresholds. Our system will provide fully personalized recommendations that consider each individual's walking dynamics to determine comfortable and safe thresholds that minimize pain and enable the person to become progressively more active. 

2. Recommendations are based on the assumption that people will be engaging in formal exercise (e.g., going for a walk or using a treadmill). Lack of time, for all demographic groups, has been consistently identified as a barrier to exercise. Our system embeds recommended levels of walking into a person's travel routine so that physical activity is a normal part of daily life. Importantly, our system identifies opportunities for travel behavior change (e.g., from car to public transport) that result in minimal or no time investment to achieved desired/recommended physical activity goals - the gift of time to be healthy. 

Outside of the clinical setting, adherence to physical activity recommendations are abysmal. Recent studies prove that personalized social aspect of habits have promise to induce behavior change (read: Social sensing: Obesity, unhealthy eating and exercise in face-to-face networks https://dspace.mit.edu/handle/1721.1/65376). 

Importantly, smartphone ownership in older adults continues to increase around the world - it has quadrupled in the United States (with around 50% of older adults owning a smartphone) and is now over 75% in Australia.

We are currently serving 150 people:

• 100 students/faculty/staff of the University of Adelaide who have a range of mobility, hearing, vision and other impairments that limit their mobility. We are using our technology to learn how they move to/from/around the campus, and identify evidence-based solutions to meet their mobility needs.
• 50 people through our validation trial with Yarra Trams (world’s largest tram network operated by Keolis Downer, a French-Australian partnership and the Victorian Department of Transport) in Melbourne, Australia.

In one year, we expect to be serving around 5,000 people with our technology – we expect that around 500-1,000 of them will be osteoarthritis patients at the Motion Analysis Lab of the Harvard University Medical School and the Florey Institute of Neuroscience and Mental Health (the latter will be patients with Type 2 diabetes).

In five years, we expect to be serving around 1-5 million people with our technology (across health, planning and transport) – we expect that around 100,000-500,000 of them will be patients with mobility impairment and chronic diseases (including osteoarthritis) in healthcare organizations in the United States, Australia, Kuwait and other countries around the world. 

Our technology has a wide range of applications in planning and transport that will help establish smart, resilient and sustainable cities. By promoting active travel for everyday activities, our technology will not only directly impact people’s health, but will also promote reductions in air pollution, greenhouse gas emissions, congestion, road injuries/fatalities, benefiting the lives of millions of people around the world.  

Our primary goals are:

• Technology development – the 1-2 year goal is to complete development of all technology components and test an integrated platform.
  • Biomechanical detection system with the Harvard University Medical School
  • Integration of activity detection and biomechanical detection systems
  • Context detection and dynamic decision-making AI system to infer contextual and environmental information (e.g., public transport crowding, heel-height and injury) and adjust recommendations in real-time.
  • After vetting A Piedi in collaboration with our research/technology/test partners, we will scale globally. Worldwide, over half a billion people suffer from osteoarthritis; increasing physical activity levels will have significant health benefits.

·        Grow and scale our solution to improve the lives of millions of people around the world – years 2-5

Our technology has diverse applications and will contribute to addressing multiple challenges:

• Physical Inactivity – 4th leading risk factor for global mortality. U$117 billion in annual costs in the US alone.
• Congestion – U$305 billion in US & over €100 billion in Europe.
• Air pollution – 4.2 million deaths p.a. (cost U$5 trillion). WHO: 90% of people live in places with bad air quality.
• Climate change – devastating global health/environmental/economic impacts.
• Road Safety – 8th cause of death across all ages and #1 for children.

It is well-known that poorly planned/car-dependent cities are one of the primary contributors to these and many other global problems. Health/policy/infrastructure/planning decisions are still largely based on inaccurate/unreliable/incomplete/untimely data. Our system will provide the data needed to promote the establishment of healthy/resilient/sustainable cities. 

Three barriers are key focus points for A Piedi:

• Walking as a proposed solution to chronic diseases of epidemic and catastrophic proportions may seem archaic, insufficient and lacking ‘wow’ factor – this may limit our ability to secure grant funding and investment.
• Ability to create a viable business model for the health applications of our technology that enables the establishment of a sustainable business that allows us to serve every human being in the world and prevent and manage chronic diseases (including osteoarthritis). One of the primary challenges is that close to 90% of public and private sector spending on health is for medical services. This, despite the fact that lifestyle and behavior are the most important contributing factors to individual and population health.
• Privacy concerns may limit widespread implementation. The activity detection system we have developed creates highly precise and detailed movement pattern data about what people do, when and where they do it, and how they travel. This may deter some people from using our technology. 

The following are our plans to overcome those barriers:

• We are firm believers that ‘simplicity is the ultimate sophistication’ (Leonardo da Vinci) – our technology utilizes highly advanced artificial intelligence techniques and the latest knowledge in wearable sensors, biomechanics and robotics to provide personalized walking recommendations to patients with osteoarthritis and other chronic diseases. Participation in forums like MIT solve will enable us to spread our message and garner the support we seek.
• ‘Convincing’ healthcare organizations to invest on solutions that focus on prevention and management of chronic conditions (rather than on medical services and treatment technologies) will continue to be challenging. However, there is growing evidence that healthcare organizations (including insurers and wellness entities) have identified explicit value in promoting healthier lifestyles and behaviors. We are confident that our success in enhancing physical activity levels for those patients most in need of exercise (and who also cost the most to healthcare organizations) will attract global attention and result in sustainable business activity. The transport and planning applications of our activity detection system will also provide a strong foundation of revenue to support the development of the health side of our business.
• Our system will provide valuable benefits to organizations and end users. As such, privacy-fueled resistance will be mitigated. There is growing evidence that people are willing to sacrifice privacy when they trust the organization managing their information and when they feel that there is a clear individual benefit.

N/A

Our company's team includes the following key members:

• Shaikha AlOthman | Founder
• Jose Mantilla, SM (Course 1) | Founder
• Diego Jara, PhD | data science adviser
• Rodrigo Mazorra, PhD | behavior change adviser
• Luciana Ledesma | commercialization adviser

Our team combines a peer-to-peer technology specialist (Shaikha), a mobility-health connection pioneer (Jose) and an adviser who specializes in introducing technologies for patient mobility gains for osteoarthritis patients (Paolo).

Shaikha is an award-winning MIT-based health entrepreneur pioneering the peer-to-peer technology for the aging population in the US. Her solution has won the MIT Hacking Medicine Grand Prize and Weill Cornell Medicine awards, and serves family caregivers of elderly loved ones. Shaikha has been implementing the project under the mentorship of MIT in a National Science Foundation program, and piloted the solution in Burke Rehabilitation Hospital in New York. Her work has been identified by the U.S. Department of State amongst 1,200 hand-picked entrepreneurs worldwide.

Jose has a Master of Science degree in environmental engineering from the Massachusetts Institute of Technology and over 20 years of international government and consulting experience. He provides strategic guidance to public and private sector clients on transportation policy and infrastructure issues. He is pioneering the integration of health, transportation, planning and human behavior sciences to create smart, sustainable and resilient communities, tackle climate change and enhance individual health.

Paolo is the Director of the Motion Analysis Lab at Spaulding Rehabilitation Hospital. An internationally-known rehabilitation engineer, he has consolidated the lab’s preeminent position in biomechanics and rehabilitation. He is Assistant Professor in the Department of Physical Medicine and Rehabilitation at Harvard Medical School and is a member of the Affiliated Faculty of the Harvard–MIT Division of Health Sciences and Technology in Cambridge.

Our research partners include the Motion Analysis Laboratory at the Spaulding Rehabilitation Hospital of the Harvard University Medical School, the Florey Institute of Neuroscience and Mental Health and Swinburne University of Technology.

Our technical partners include leading data science and technology companies in Belgium (Sentiance), Colombia (Quantil) and Australia (MoWorks).

We are currently delivering the following projects:

• Campus accessibility strategy for the University of Adelaide. We are applying our technology to understand arrival/departure/on-campus movement patterns for students/staff/visitors with a range of disabilities (e.g., mobility, vision and hearing impairment). The primary objective is to develop evidence-based solutions to promote safe, convenient and comfortable access to the university facilities for people of all ages and abilities.   
• Keolis (French-based public transport operator) and the Victorian Department of Transport. We are applying our technology to determine its potential for understanding citywide movement patterns that can inform the development of evidence-based infrastructure, planning and policy solutions.

Our technology integrates with existing systems – we embed our intelligence into existing smartphone apps and feed the outputs to our data analytics platform. We are fusing areas of knowledge that operate in isolation – world’s first application of AI that integrates health/transportation/planning/human-behavior sciences. Our product does not exist in the marketplace.

Beneficiaries are people who want to be more physically active and who want to travel in a more intelligent, efficient and environmentally friendly manner.

Our primary customers are health insurers and healthcare providers – they want our technology as it determines how much people can move and provides personalized, safe, realistic and desirable physical activity recommendations that are commensurate with their capabilities and are embedded into their daily routines. Health organizations incur significant costs directly attributed to physical inactivity and associated chronic diseases.  We help promote physical activity, which could result in significant savings. We offer them the ability to profile risk of members by evaluating movement/physical activity and personalized intervention to promote incremental/meaningful behavior change, thereby lowering costs and promoting healthier outcomes for their customers. 

Since our technology creates fine-gran and highly precise movement pattern data, any organization that is interested in understanding how people travel is a potential customer. This includes public/private transport operators, government planning/transport agencies and property/infrastructure developers/operators – they want our technology as it creates essential data about people’s movement in cities that has never before been available and has been cost-prohibitive to acquire. 

The primary mechanisms to fund our work will be a combination of selling products/services and raising investment capital. We will continue to selectively apply for research and development grants in the United States and Australia. We are fully confident that our revenue stream will fully cover our expected expenses and result in financial sustainability and profits. We will reinvest a significant proportion of our profits into further development of our technology.  

We offer various licensing models to clients in the health and transport/planning sectors. In health, we will offer a per-user service package or licensing agreements with insurance companies (health, life, accident and corporate), wellness organisations and corporations – all share a particular interest in promoting physical activity to: lower claim costs, reduce premium expenditure, increase life expectancy, enhance general health/well-being, reduce absenteeism and/or increase workforce productivity.

In transport/planning, we will primarily offer use of our data analytics platform on a per user per month license-fee basis. Our technology provides far superior movement pattern data (i.e., real-time, revealed, fine-grain and over much longer/continuous collection periods) at a fraction of the cost of current methods. Transport operators, infrastructure providers, government agencies and property developers collectively spend 100s of millions of US$ per annum collecting poor and incomplete urban data.

Wherever possible (given privacy and other considerations), we will retain partial data ownership (in aggregated and anonymized format). This will enable us to generate additional revenue with municipal/state/federal agencies and private property/infrastructure developers interested in mobility insights.  

As an early-stage venture, we are applying to Solve to expand our network of potential collaborators, formal partners, customers and investors. We are convinced that no single organization can claim – or should claim – to have ‘all the answers’. As such, even though we are proud of our collaboration with Harvard University, we are fully cognizant of the immense value of joining an MIT-backed community. Our team leaders have experienced firsthand the strength of MIT in the research and innovation arenas, and we look forward to the opportunity of joining Solve to accelerate our work, validate our impact and business model, and scale our solution.

In particular, we are interested in being part of a peer network with organizations who are involved in the development of healthy city solutions, in order to exchange learnings, advice, support and networks. We are keen to meet with potential partners in the areas of digital health, active travel, smart cities and behavior change science. Based on the requirements to meet our growth and scaling goals, Solve's connections will be invaluable to ensuring access to a broader pool of technical, scientific, strategic and financial resources. 

Given the magnitude of the physical inactivity epidemic and the lack of personalized solutions for those individuals most in need of help, our solution has the potential to scale rapidly at the international level. We believe that being part of the Solve community will enable us to build the partnerships needed with players in industry, nonprofits, government and academia. 

N/A

Our solution has direct applications for a wide range of industries in the health sector that have an explicit interest in promoting physical activity for people suffering from osteoarthritis. We are particularly interested in expanding our solution to other population groups – this would open access to a virtually limitless untapped potential for increasing physical activity as part of daily travel. Potential partners include health insurers, healthcare providers, digital health businesses and well-being organizations.

Promoting physical activity, however, is also of interest to private entities and public agencies in the planning and transport sectors. Reliance on motorized modes of travel is not only strongly associated with declining physical activity levels worldwide, but is also at the heart of a wide range of local and global problems, including road crashes (#1 killer for children worldwide and #8 across all age groups), congestion (crippling productivity), air pollution (90% of people in the world live in places with unhealthy air, responsible for 4.2 million deaths yearly) and climate change (devastating effects already being felt in communities and ecosystems globally). Other ‘obvious’ potential partners thus include municipal/state/federal agencies and transit operators.

Within  this context, we are keen to work with the Solve community to explore partnerships that enable us to make the knowledge/insights that are currently only available to people at advanced rehabilitation facilities (when it is already arguably too late) to people ‘on the street’ with chronic conditions. Our only requirement: working with people/organizations who care as deeply about these issues as we do. 

Our core R&D areas focus on the use of advanced Artificial Intelligence methods. The Activity Detection system that we have already developed (and that we will use as part of our solution to identify what, when, where, how and why people travel) uses unsupervised learning techniques to detect and predict travel patterns in real-time. This system includes adaptation of the DeepCity learning framework to map destinations (even in the presence of noisy location estimates) based on an individual’s past behavior.

The Biomechanical Detection system that we are developing (with the Motion Analysis Laboratory at the Spaulding Rehabilitation Hospital of the Harvard University Medical School) will use mobile sensor data and self-adjusting algorithms (based on Random Forest classifiers) to analyze an individual’s walking cadence/gait patterns. Artificial Intelligence techniques will be used to determine fully personalized physical activity recommendations based on ‘pain thresholds’ captured through analysis of smartphone sensor data.

The integration of the two systems will rely on online evolutionary learning techniques to concurrently deliver effective recommendations to individuals and increase the genetic diversity of the algorithms. We will tailor online learning algorithms for responsive and privacy-preserving adaptation of physical activity recommendations to changes in environmental and personal contexts.

We will utilize the prize to advance the development of the Biomechanical Detection system and the integration with the Activity Detection system. This will enable us to have a world’s first Artificial Intelligence platform that provides personalized physical activity recommendations that are commensurate with individual capabilities and embedded into daily travel routines. 

N/A

N/A

Osteoarthritis is a degenerative joint disease, which mainly affects the articular cartilage. It is associated with aging and will most likely affect the joints that have been continually stressed throughout the years including the knees, hips, fingers, and lower spine region. The World Health Organisation has identified osteoarthritis as one of the most disabling diseases in developed countries and the single most common disability cause in older adults. Worldwide estimates are that 1 in 10 men and 1 in 5 women aged 60 years and older have symptomatic osteoarthritis. Unfortunately, 80% of people affected by osteoarthritis will suffer movement limitations which will exacerbate the onset of chronic diseases, depression and other mental health illnesses, and accelerate cognitive deterioration.

The solution we are developing directly targets osteoarthritis, through the use of innovative science and technology to improve the quality of life for all people, particularly women (as they are more likely to develop osteoarthritis throughout their lifetimes and given the much higher prevalence of the disease in older women). Our solution can be easily transferred to women with other chronic diseases, such as gestational diabetes, and more broadly to the general population. As such, it will promote improvements to the quality of life of women and girls worldwide by providing personalized, medically safe, achievable and realistic recommendations for physical activity that are commensurate with individual capabilities and embedded into daily travel routines.  

Our core R&D areas focus on the use of advanced Artificial Intelligence methods to create a differential and disruptive impact for our economy and society.

·        Activity Detection system – already developed (we will use to identify what, when, where, how and why people travel). Uses unsupervised learning techniques to detect and predict travel patterns in real-time. Includes adaptation of the DeepCity learning framework.

·        The Biomechanical Detection system – developing (with the Harvard University Medical School). Will use mobile sensor data and self-adjusting algorithms (based on Random Forest classifiers) to analyze an individual’s walking cadence/gait patterns. AI techniques will be used to determine fully personalized physical activity recommendations based on ‘pain thresholds’ captured through analysis of smartphone sensor data.

·        The integration of the two systems will rely on online evolutionary learning techniques to concurrently deliver effective recommendations to individuals and increase the genetic diversity of the algorithms.

Our system will place primacy on individual privacy protection, with anonymization of travel patterns, de-identification (through removal of direct identifiers and alteration of quasi-identifiers), and layered security and control of the collected data. Protecting personal data is of the utmost importance to our work and we take all necessary precautions to safeguard private data and process it only in compliance with the EU General Data Protection Regulation. The guiding principles of our data privacy approach are transparency, lawful processing, data minimization and data security. 

The World Health Organisation has identified osteoarthritis as one of the most disabling diseases in developed countries and the single most common disability cause in older adults. Worldwide estimates are that and 1 in 5 women (as opposed to 1 in 10 men) aged 60 years and older have symptomatic osteoarthritis. Also, women tend to be plagued with a more severe modality of osteoarthritis than men do. Unfortunately, 80% of people affected by osteoarthritis will suffer movement limitations which will exacerbate the onset of chronic diseases, depression and other mental health illnesses, and accelerate cognitive deterioration.

The solution we are developing directly targets osteoarthritis, through the use of innovative science and technology to improve the quality of life for all people, particularly women (as they are more likely to develop osteoarthritis throughout their lifetimes and given the much higher prevalence of the disease in older women). Our solution can be easily transferred to women with other chronic diseases, such as gestational diabetes, and more broadly to the general population. As such, it will promote improvements to the quality of life of women and girls worldwide by providing personalized, medically safe, achievable and realistic recommendations for physical activity that are commensurate with individual capabilities and embedded into daily travel routines.  

Importantly, our solution includes carers and friends/family in the conversation; the social engagement – often overlooked by technical solutions – is at the heart of our joint degeneration mitigation mechanism.