Weakness & Improvement

Weakness

There is a few weakpoints we discovered by doing this project:

The data we get during the mesurement is continuous and not only when the finger is on the pulse sensor. Sometimes we might obtain some wrong/impossible values.

In addition, the pulse sensor is not very precise, and it could be also improved if we used a better one or isolate it from external interference in a better way. In addition, it needs about 15 seconds to start obtaining more stable measurements.

Secondly, we are using the free version of ThingSpeak, so the values we get are updating only every 15 seconds which is pretty slow, which makes real-time monitorization difficult and not that suited for a real-case scenario.

Another key point is that in really the accelerometer should be placed on the wrist or on the chest, however due to the large size of the board the person in question currently needs to move it with the hands to simulate movement.

While the app offers the possibility to create a new user account, as no external database has been designed, the demo assumes that there is always one same patient. Therefore, in order to change the “username” and “password” it is necessary that these be modified from the Android Studio code.

Improvement

Like said on the weakness part, better components can help to improve the quality of the values we obtain during the project.

Firstly, it would be necessary to preprocess heart rate data more accurately so that data is only taken into account once it has more or less stabilised.

As for the app, we can also add a real database in which we save data of the patients and doctors using our app and change/retrieve data of each

for a specific ID, to make the system more secure and realistic.

We could also create a specific login so doctors and patients can have their own account. For example when the doctor logins, he will be able to see all the registered patient.

Another add-on could be to be able to change the telephone numbers we set initially, which is related to the lack of a database.

The app is so far only downloable from the github repository, although in the future it would be interesting to make it available on Google Play for easier download.

The SMS activation conditions have been estimated generally. Preferably the doctor or familiar should decide when they would like to be notified, as for instance, depending on whether the patient tends to do physical activity or not, "normal" conditions may differ.

·Lastly the overall design of the solution could be improved, both to improve functionality and also aesthetically, among other things, by reducing the size.

Reflection

The main problem we had was establishing connection between analogue sensors and wifi-module Node-MCU. In the end it was decided to communicate them using Arduino’s serial connection, however the data reception from the sensors and readings from Node-MCU did not match. This was solved by encoding the accelerometer and sensor data was send to the serial.

In general we obtained to design a