| |
|
For more detailed list of current projects, please visit our group's web site:
GROUP WEB SITE LINK
|
|
|
Point of Care (POC) Diagnostics and Machine Learning
With my student group at MIT, we design mobile devices and mobile apps to support health applications, including diagnostic and therapeutic tools. Although the immediate need is for global health in rural clinics around the world, there is certainly an emerging need in developed contries as well for consumer health. This work includes research in three main areas:
- Low-cost Devices -- I design clever low-cost electronic and optical devices that interface to a mobile phone and can be used for diagnostic support or disease screening (e.g. anemia) in resource-poor settings.
- Machine Learning -- My group develops advanced algorithms that go beyond simple data collection to provide diagnostic and valuable clinical decision support for the field doctor or health worker. This work includes advances signal processing, blind source separation algorithms, and probabilistic inference models that are also efficient and can run in real-time on a mobile phone or tablet.
- User Interface Design -- Making diagnostic tools available to a wider community required completely rethinking the user interface design and using advanced techniques such as augmented reality and imaging technologies designed to simplify clinical use and transcend language barriers.
|
|
|
Wearable Sensors for mHealth
Wearable technologies are changing the way we think about healthcare and enabling lower cost personal preventative health. Future wearable sensors will never need batteries or maintenance and can be comfortabely worn as clothing or jewelry. Despite numerous commercial products, most are not appropriate for clinical research and contain little intelligence.
Research in wearable sensors is interdisciplinary and includes wireless communications, ultra-lower power circuit design, as well as knowledge of clinicla physiology, biology and chemistry. Wearable sensors measure not only physiology (heart rate, skin conductance, etc.) but also include information about environment parameters (air quality, sound/light levels) which affect our health.
In addition to sensors, our electronics modules contain low-power radio which stream data to the cloud. This enable us to perform longitudinal data analysis on a remote server, which important for clinical studies and outpatient care betwen doctor's visits.
|
|
|
Global Health: Maternal, Newborn, and Child Health (MNCH)
Among the eight international development "Millennium Goals" established by the United Nations in 2000, two main goals are reducing child mortality and improving maternal health. I work with the Global Health Division at Massachusetts General Hospital as well as D-Lab at MIT to develop and deploy a variety of technology tools that can improve the lives of pregnant mothers, newborn infants, and young children. These tools range from simple health education to mobile diagnostic tools for use by health workers in rural primary clinics. I would like to also acknowledge the MIT MISTI-India program for providing some seed grant funds to help catalyze some of these projects and enable me to travel to many locations in the field.
|
|
|
Mobile Psychiatry and Behavioral Therapy
Mobile phones are powerful tools that can deliver therapeutic content and enable connection to our support network. We are exploring ways that these Mobile Health platforms can be used therapeutically to deliver health interventions. Phones can be used as a vehicle for cognitive behavioral therapy or as an adjunct to clinical therapy and psychopharmacological treatments.This work exists on multiple-levels:
- Psychophysiology and Neuroscience --I collaborate with neuroscientists and psychiatrists, working with both animal models and humans, to better understand the neurological and psychophysiological underpinnings of addiction and pathological behaviors.
- Behavioral Monitoring -- Using sensors that enable ambulatory measurement of autonomic nervous system activity (SNS/PNS), we can investigate mental health disorders and co-morbidities, including substance abuse, anxiety disorders, sleep disorders, and depression in particular. My work also includes studies with homeless populations and military veterans.
- Technological Interventions -- I design software platforms that provide a range of mobile tools, ranging from simple text messages to multimedia videos that can be administered via the web and delivered via mobile phones.
|
|
|
Environmental and Ecological Monitoring
My research takes a more holistic view on environmental monitoring, to include social as well as physical factors:
- Physical Environment -- as part of architecture and city design, many aspects of our physical environment, including air pollution, light color, and sound, have each been shown to have an impact on our physiology and health.
- Social Environment -- The frequency, mode, and affect of our social interactions (live conversation vs phone vs online) are important influences on our health. These factors can be mediated through design of our living spaces and adopting healthy behaviors.
- Differential Vulnerability -- We are discovering significant differences in the way each person is affected by the environment. The impact of the environment on our health is highly idiosyncratic, depending on individual factors such as genetic predisposition, co-morbidities, and emotional mood.
|
|
|
ICT4D: Information and Communication Technologies for Development
I've devoted a large part of my life to creating technologies to help basic human problems in developing countries. Designing technology within the constraints of extreme affordability and minimalist design is surprisingly challenging and often leads to completely new solutions to traditional engineering problems.
"HELLO WORLD: Technology Missionaries to a Crowded Planet" -- This is my term paper from my PhD seminar class (1997), when I started exploring technologies for developing countries. Here is an excerpt from Page 1:
"Technology inevitably advances. Moore's Law states that computers double in speed every two years, and thus far this decade, it has held true. Unfortunately, there is no such Moore's Law for describing progress in solving classic human problems such as disease, hunger, population, pollution, and illiteracy. In developed regions, computers, wireless links, and other information technologies have increased our productivity and improved our overall quality of life; however, most of the world remains alienated from the benefits of such technology. Should technological progress be measured simply by the number of transistors that we can squeeze onto a piece of silicon, or should it be measured in terms of the value it has to our quality of life?"
MIT Courses I have co-founded and helped teach:
EC.712/782, "D-Lab-ICT: Information and Communication Technologies for Developing Counties" (Fall'09,'10,'11) Class exploring the broad range of ICT technologies applied to specific problems in the developing world.
NextLab I, "Designing Mobile Technologies for Developing Countries" (Spring, '08, Fall '08) Class focused on Mobile phone platforms. Part of Next Billion Network
MAS968, "IT for Developing Counties" (a.k.a. Wind-up Browser)(Spring '99) (with Saul Griffith).
|
|
|
Digital Agriculture: Farm in the Cloud
As more people migrate to cities, and clean soil/water become more scarce, there is a need to explore alternative methods for locally producing food. While growing plants is deceivingly simple, and alternative methods have existed for decades, fully automated and reproducible production of food at the local scale has not yet been possible. Within the Smart Cities group at MIT, and CityFarm project, my research includes work on two main levels:
- Agriculture Sensors --I develop low-power or self-powered sensors which are low-cost and can be scaled for use by communities for monitoring plant growth and conditions. Counpled with environmental controls, this forms a feedback loop for growing plants.
- Network Architecture -- While I work with farms that range in cost from $100 to $60,000, the most important aspect of our research is the development of a network architecture or API that lives in a virtual machine in the cloud, which we call "Open agriculture". We are developing server algorithms that process the large amounts of data from our sensors and then automatically control devices in our farm that control the lighting, nutrients, and environment.
|
|
|
Low-cost Self-powered Sensors and Data-Loggers
I design ultra-low-power data loggers that can record sensor information and can interface to ubiquitous mobile phones. Here are some sample applications of this technology:
- Cooking stove use -- Cooking stoves are a important health concern. Temperature data loggers powered by the heat of the stove itself can be used to monitor how often stoves are used as well as usage patterns.
- Medication compliance -- We can measure how often certain medication is used, such as pill bottles or inhalers.
- Water filters/irrigation -- By connecting the data logger device to a valve, we can monitor water usage for drinking or irrigation on a farm.
|
|
|
Wireless Sensing and RFID
Wireless sensing and RFID is an interdisciplinary field ecompassing antenna design, radio-frequency design, and sensor materials science, as well as protocol/algorithm development and coding strategies. Many of these elements are interdependent, and thus low-cost RFID and wireless sensors are by definition cross-layer designs. Early work on wireless sensing and RFID at the Media Lab took place in the mid-1990's beginning with the research consortium called Things That Think For more information on this technology, you can request from MIT a copy of my Master's or PhD thesis devoted to low-cost electromagnetic tagging. Click here to see a short online overview of chipless tag technology I wrote in 1999; although somewhat outdated, it still contains a useful survey of RFID and chipless technologies.
In March 2000, I founded a boutique RFID design company called TagSense, Inc. which designs and manufactures next-generation RFID systems for various clients.
|
|
|
Networked Environmental Sensors
I create low-cost sensors for monitoring the environment (air, soil, water). Some of my first projects included weather probes for Mt. Everest and soil sensors for poor farmers in central america. Depending on the focus of the project, we sometimes use environmental data for geophysical reseasch, farmer livelihood, health (e.g. asthma), or education.
Below are past and current projects that involve environmental sensing:
|
|
|
Plant Instrumentation and Cybernetics
Plants are very common in our world and and contain a vast amount of information. Although there are open debates about the intelligence of plants, it is undeniable that plants have a great ablity to sense and respond to their environment. The electrophysiology of plants has sparked interest since the late 1800s, but this topic has not been explored recently in the context of modern information technology and electronic devices. We are building novel electronic and optical sensors that measure the physiological activity of living plants, and explore their use as low-cost sensors and thought-provoking educational tools for children and museum exhibits. There is also increasing interest in the study of plant communications.
|
|
|
A Printer that Uses No Ink or Consumables
As a design study in sustainability, I set out to build a printer that uses no consumables (ink, paper, disposable batteries). Traditionally, printers are one of the worst examples of waste. The completed design comprises reusable photochromic materials, low-power motors, and a solar power battery charger.
Below are related links:
|
|
|
Store-and-Forward Wireless Networks
802.11 WIRELESS ON VEHICLES I have developed and commercialized a simple store-and-forward concept for rural "last-mile" connectivity that was inspired by the satellite system we used for our Mt. Everest data link. In 2001, I joined forces with other students who interested in this topic and entered this idea in the 2002 MIT entrepeneurship business competition. Our team made it to the semifinals. In Fall 2002, I co-founded a new company with my business partner Amir Hasson, called FIRST MILE SOLUTIONS, LLC . We teamed up with Prof. Sandy Pentland to write a paper titled "DakNet: Rethinking Connectivity in Developing Nations" which became the cover story of the Jan. 2004 issue of IEEE Computer Magazine.
|
|
|
Simple Technology Design
I value simple design, and simple technology. Doing more with less. Digital technology and fancy tools have made us (inventors) lazy. Over the past century, people like J.C. Bose are a great source of inspiration. Working in austere Calcutta, India in the late 1800's, Bose was not only a pioneer in microwave and wireless devices, but he also was one of the first people to connect electronic devices to living things. Since 2000, I have been forming a design and educational philosophy to help develop "Simple Technology Design" as a new discipline.
|
|
|
Electromagnetic Safety Sensors for Automobiles
There are several thousand bicycling accidents yearly (some fatal) in which a cyclist crashes into the door of a parked car which is opened suddenly. Such accidents are very common here in the Boston area, where few bike lanes exists, on-street parking is common, and the streets are narrow. We have developed a side-mounted automobile safety sensor which detects the oncoming cyclist and sounds an alarm to the driver. Alternatively, the alarm signal from the sensor can be used to activate the car's door lock or flash the rear-view lights or sound the horn to warn the cyclist as well. The sensor is based on a simple microwave doppler radar circuit to which we have added a custom designed narrow beam helical antenna which focuses the sensing zone to the volume immediately adjacent to the car door without being affected by the driver's head movement or other oncoming traffic. Although the microwave circuit and antenna design is non-trivial, the entire sensor assembly can be easily copied and manufactured for a cost of approximately $20.
It is worth noting that our lab has also developed other automobile safety sensors for air bags based on capacitive sensor technology. This project was led by Josh Smith and Prof. Neil Gershenfeld. These sensors have now been commericalized by Honda/Acura and Ford, and are now being used in milions of cars worldwide.
|
|
|
Medication Compliance
In terms of money, medication compliance is often listed as the biggest problem in healthcare. When a medication treatment fails, it is not known if it failed because it was the wrong medication or because the patient did not take the medication properly. Over the past 20 years, a wide variety of inventions have been created to address this problem, with mixed success. Below is a system I designed with Paul Yarin in 1998 (see video). Later versions were patented with our sponsor Becton Dickinson.
The Sensepad Medical Monitoring system (early prototype 1998 shown in photo on left). Click on the image to view video.
To my knowledge, this was also the first demonstration at MIT of a smart shelf employing RFID-tagged objects (1998).
Download the Video
Our early medication compliance systems used embedded dial-up modems and pager interface. However, current concepts make use of mobile phone technologies. An example of current approaches to medication compliance done by or collaborators in D-Lab at MIT can be found here. |
|
|
Sensor Platform for Remote Monitoring of Infants
For home use as well a clinical applications, there is a need for portable monitoring of infants. Several of our wireless sensor platforms have been applied to this application. An early project (with Gili Weinberg and Sum-Lin Gan) was an infant monitoring system using fabric-based capacitive sensors with tangible and ambient displays (video clip available below). More recently, my research includes a variety of wearable wireless sensors that measure physiology, such as baby electrodermal activity and heart rate:
Musical toys for babies (with Tod Machover's Group)
Download the Video |
|
|
Nuclear Magnetic Resonance (NMR)
We A more elegant approach to electromagnetic sensing of objects can be done via NMR. The NMR signature of a given material can be detected in a very similar manner as detecting any other magnetic resonance. The most common uses of NMR are in chemistry and medical imaging (you can see an MRI scan of my head on my personal web page). We (i.e. Neil Gershenfeld) are interested in using NMR to do logic computation in materials, but we are also interested in making cheap simple NMR readers to do crude identification of materials which would be useful for a variety of applications (security, surveillance, chemical sensing).
I adapted a table-top NMR system I had used in an undergrad Physics lab to work with a permanent magnet and miniature probe and set it up at the Media Lab. You can see our lab's first working NMR system here (June 1997). Since then, fellow students Yael Maguire and Jason Taylor have design and built high-performance table top NMR systems which can be applied to the field of quantum computing.
My personal interest in NMR is not for computing, but rather for interesting sensing applications. A low-cost NMR unit would enable, for example, refrigerators and appliances which know if your milk has spoiled or how much fat is in the dinner you are cooking tonight.
For a primer on NMR and MRI, there are many good references online, such as the following:
Tutorial on NMR (courtesy Rochester Institute of Technology) |
|
|
Mount Everest Weather Probe 1998
In 1999, I was one of the members of the team which succeeded in making 2 working remote weather probes for Mt. Everest. These probes transmit data every 2 hours via satellite and is then automatically posted on the web. These probes transmitted data daily since May 5, 1998, and lasted until late August when their batteries ran out. For more info on the weather probes, click here.
|
|
|
Mt. Everest Weather Probes 1999
In 1999, we participated in a second Mt. Everest expedition. The weather probes we used the second year were similar, but with longer battery life and an improved wind speed sensor. The Everest weather probes were successfully deployed at several locations near the summit of Mt. Everest during the first week in May 1999 by climber Pete Athans and his crew. One of the weather probes continued transmitting data for eight and a half months until early December 1999. That year we were sponsored by the Weather Channel.
|
|
|
Force Sensing and Actuation for Human Computer Interface Applications
A goal in HCI research is to make technology that literally feels good and make interfaces that are comfortable and useful. It is sometimes desirable to relay information in the form of tactile or force interactions. New classes of materials, such as smart materials and magnetic materials, provide new or improved methods for force transduction. Force can be used either as an output for haptics, as an input for sensing, or as a novel -- but perhaps inefficient -- means of generating energy. Some early concepts are described in a simple survey paper on materials, mechanisms, and media applications of force transduction which appeared in a special issue of IBM Systems Journal.. |
|
|
Materials Characterization
Over the years, we have used a variety of materials characterization techniques to explore new types of sensors and also better understand the materials we are sensing. Some of these techniques were developed during my time working at the US Air Force Materials Lab.
|
|
|
|
