DISCLAIMER: This article is about version 1.0 of our product. We’re constantly iterating to improve our product.
Before I even start, watch this video if you’re here just for a quick understanding of what we built.
If you’re reading this right now, you’ve probably had some problems with the healthcare system. Maybe the commute to the clinic, the wait time, or maybe even the germy waiting areas. However, for over half the world’s population who don’t have access to quality healthcare, it’s 10x worse!
Over 65% of physicians worldwide feel overworked, and that’s including physicians from the first world and third world countries.
And a lot of the time it’s just for a quick and simple diagnosis that you could probably do yourself.
Unfortunately, if you’re someone who lives in a rural or in an impoverished area, you don’t have the luxury of accessible clinics or hospitals. Right now your best bet would probably be to rely on online doctors or also known as telemedicine.
But telemedicine has so many limitations mainly because online doctors don’t have access to your body’s metrics to give you an accurate diagnosis as a clinic would.
My friend Sigil and I thought about what kind of healthcare system we wanted to exist in 15 years and how we could build it today.
Imagine if we had a way to easily deploy access to healthcare in third world countries, rural areas, and, medical desserts.
The major issues we addressed were a better system for diagnosis for online doctors and physical doctors and the accuracy of the basic diagnosis.
What we built was a cost-effective portable device that uses IoT biosensors and cloud-based machine learning to help online doctors.
The Hardware, everything from biosensors to the raspberry pi
It all runs off a raspberry pi 3 which is basically our CPU and the central component of our application. From our raspberry pi 3, we connect it with an LCD screen.
On the LCD screen, you can interface using a touch stylus pen to answer basic questions about how you’re feeling or symptoms to help assign a specific sensor for you to use. On the LCD screen, it would also connect you with an online doctor if needed.
From the raspberry pi 3, all sensors which might be needed are connected. So you’d just use a certain sensor following a simple tutorial from the LCD screen and all your data would be automatically synced in a database for you and your physical or online doctor.
Here’s a breakdown of our hardware components for our first version of the product.
- A Camera and LED Light — This is used for taking pictures of parts of the body like the ear and tongue, which give a lot of data about certain ear infections or other bodily infections and illnesses
- Temperature Sensor — It detects temperature in certain parts of the body
- Stethoscope Electrocardiogram Sensor (ECG) — a sensor to collect the heart of the patient (cause not everyone has a smartwatch)
- Pulse Oximeter — To detect blood pressure within the body
- Raspberry Pi 3 — To be the centralize CPU to connect the sensors altogether
- LCD Screen — To communicate with online doctors and walk through using specific sensors
- Microphone — Allows you to communicate with a remote online doctor
The Camera and LED
The camera would be used for specific functions like taking pictures of the ear and the tongue if needed. Having accurate and precise pictures of places like the ear and tongue it would make it much easier for doctors to give a more accurate diagnosis of certain basic clinical illnesses.
Stethoscope with integrated ECG
If the doctor needs to understand your heartbeat then you’d need to use a portion of the stethoscope. Where the bottom of the stethoscope would be used to get the heartbeat and it would play the heartbeat sound and speed into the speaker/mic of the device.
So you could also hear your heartbeat just like a doctor would.
The temperature sensor detects the temperature in any part of the body and is especially useful for detecting illnesses like the common cold. It takes the temperature, stores it in a raspberry pi 3 then sends that to the same database your clinical doctor would have.
Raspberry Pi 3 and LCD Screen
If you’re thinking of it in the software sense of it — you can think of the LCD screen as the front end and the raspberry pi 3 as the backend. The raspberry pi 3 does the work that often isn’t seen by the user like sending the sensor data to the database and mainly allows for the functionality of the LCD screen.
While the LCD screen is like the front end and is the method for communication with the user. That’s where you can have an interactive user experience.
The microphone allows you to communicate with an online doctor via the microphone on the device. It can take input and output making Salutem, the device, become a one-stop place to communicate with online doctors as well.
Our software is mainly going to be based on the larger telemedicine companies like Maple. Maple asks a couple of questions like how they’re feeling and symptoms that you have and directs them to a specific online doctor for that issue your facing.
We’ll use Maple’s software in order to take input on how the patients are feeling and their symptoms.
Then, they’re directed to use a certain sensor depending on how they’re feeling and factors like their symptoms.
The data from those individual sensors are automatically synced into a personal Flask database using the Internet of Things. Using the Flask database you check out the data stored easily online with good user experience.
Then using specific and individual computer vision and machine learning algorithms, using the data in the personal database a specific illness is diagnosed. And then, shared with doctors to help them with their medical diagnosis.
Salutem takes tons of different machine learning and computer vision algorithms. The range of the machine learning algorithms ranges from being as simple as machine learning for common cold diagnosis to computer vision for certain ear infection diagnosis.
It would take a condition-based on what type of sensor you’re recommended to use.
If you’re recommended to use a camera/LED, you’d be given a computer vision algorithm. If you were recommended to use any other sensor like the Stethoscope Electrocardiogram sensor, a machine learning algorithm using LSTM might be needed to give a suggested diagnosis.
The computer vision and machine learning would scrap the data from the personal database and create a diagnosis from that data in the database. And a pre-trained data for machine learning’s accuracy’s sake.
The computer vision algorithm denoised each individual image given to its original by filtering out the noise. In our case, we’d denoise an image to what it should look like if healthy. In doing so we remove the disease or the illness in the specific pictures given.
This has the ability to dramatically increase the quality of healthcare for almost everyone and especially the 400 million people in developing countries.
For people in suburban and urban areas, this has the ability to make the time they spend in a clinic much more personalized for the treatment of an illness instead of the diagnosis of the illness.
While for people in a rural and impoverished area, this has the possibility to enhance and in many cases give access to quality healthcare. That being a very low cost and accessible solution to allow people in any part of the world receive a quality diagnosis and healthcare experience.
All you really need to use Salutem is the hardware device and good internet service. From there while using the product, anyone can get quality at-home diagnosis.
If you want to learn more about what we’re doing with Salutem check out our website at https://salutem.ca/.