A research team has invented a smart handheld medical device with the aim of providing early intervention for patients with congestive heart failure.
The device, which resembles a stethoscope, is made up of an acoustic sensor connected to a smartphone. It enables early intervention by allowing patients to check for excess fluid in the lungs at home. Fluid accumulation in the lungs, which causes breathlessness, is a common symptom of congestive heart failure.
It has been invented by a research team from Nanyang Technology University in Singapore (NTU) and Tan Tock Seng Hospital (TTSH), which was led by Associate Professor Ser Wee from the School of Electrical Engineering at NTU, and worked with Dr Arul Earnest, medical statistician from TTSH who introduced Associate Professor Lee to Dr Chia Pow Li, a cardiologist at TTSH and current partner Dr David Foo.
Associate Professor Wee explained to Med-Tech Innovation News how the idea came about, saying: “In 2010, I presented in a workshop on my work on using sound-based technology and signal processing (or machine learning or AI) techniques for cardiopulmonary which involves analysing lung sounds and heart sounds to detect and monitor the medical conditions of the heart and lungs.
“The three of us held some brainstorming sessions and came up with two ideas to explore: detection of irregular heartbeat, and detection of water in the lungs, using signal processing techniques. My researchers and I did more literature and market research and concluded that there is better value in working on detecting water in the lungs as no one has done that before. I approached NTU for financial support and they gave me a seed grant to do this research with TTSH.
“Ever since I embarked on the R&D in the field of biomedical engineering in 2000, I have been talking to medical doctors to understand more about their real concerns and wish list. I have since been inspired to bring about non-invasive diagnostic and monitoring products for home use. This got me to focus my R&D on home-based medical devices, and since my expertise is in the field of audio signal processing (or machine learning or AI), I have chosen to focus my work on smart (AI) home based (non-invasive, don’t need professional to operate, can be used frequently) acoustic based diagnostic and monitoring medical devices.”
The device picks up breathing signs through a sound sensor with the sound signals sent to a server in the cloud through a mobile app over approximately 10 seconds.
Associate Professor Wee explained the challenges in developing the device, naming five in particular, saying:
“Some challenges were:
- Are there any features in the lung sounds that are distinctly different between a clear lung and one that is filled with water, and if there are, how do we find them?
- How should the Machine Learning/AI algorithm be designed to have the highest accuracy?
- How well can we pick up the weak breathing/lung sound needed for the determination of the unique feature values we identified?
- How can the product be designed to be robust for practical use (can be used by anyone where the placement of the sensor may not be exact, how long should the sensor be placed, presence of external interfering sounds)?
- How should the product be designed so that the users can use it easily and effectively, and would want to use it?”
The device comprises a sound-based sensor, a smartphone and a computer in the cloud. The sound-based sensor is a modified stethoscope.
