To prevent the spread of viruses these days the main problem is how to identify the asymptomatic cases. We propose to use acoustic signals generated by cell phones and record them. Any organic alteration caused by the virus in the ear-nose-mouth system will be identify by the microphone.
Atmospheric data provided by the space agencies will be used to eliminate disturbances from environmental conditions. An artificial intelligence algorithm will help to identify even the asymptomatic cases.
This app will be used for an early detection of potential infected people in order to help organize health policies and mitigate the spread of the virus in each region.
We are inspired by the safer future we imagine for humanity. As some viral talks around the world about the potential hazard of respiratory viruses, it was a matter of time that this situation would arrive. Now it is our task to find a smart solution to deal with coronavirus and future respiratory viruses. We consider that the best and smartest way to avoid the spread of the virus is an early detection and a rapid response to isolate the people.
Currently, the main problem is how to identify the infected people, especially those without symptoms that represent up to 80% of them. The final diagnosis includes a PCR tests, which is an accurate but expensive tool and demands get reactives for each country which are a limited resource and also represents a critical situation for those countries that don’t produce this supplies.The objective of this project is to take the existing resources to the next level in order to avoid the spread of the virus.
One of the biggest problems that we have found while working on this project was not to forget that the app will handle with real people. It was not easy how to consider the different social and personal metrics that are involved in a diagnosis like this. The diverse background of the team members was fundamental to propose a solution with which we are proud of.
NOW? In the face of an imminent economic reopening, it is essential to have the best tools to prevent a second wave of infected people, the TuneVid App can be used by government organizations to define health policies and control hotspots of virus spread. Also, it can be used by private companies such as shopping malls and locations of massive affluence of people as an in-situ control test. In case that the test results in a high value of the metric, would serve as a criterion to recommend a PCR test and isolation.
FUTURE? Tune Vid App will be able to make an early detection of other respiratory diseases and in time establish not just a potential diagnosis but suggest a grade in the evolution of the disease due to the improvement and standardization of the measurement criteria helping the medical staff to prevent collapses in the health system.
Finally, Tune Vid App will serve as a tool for global mapping in real time of new and future hotspots of disease spread.
The Tune Vid App develops 4 main stages:
1. Setting the app: In this first stage the user receives an anonymity consent, followed by an auto-calibration where the hardware phone dynamic is set to get the standard measure of the application.
Auto Calibration: Holding the phone in the air (not holding it close to ears or mouth) so what is sent by the speaker is received by the microphone, to define the acoustic transference between them in addition to the surrounding environment or air noise.
2. Organic Measurement: In a second measure, putting the cell phone next to the ear, the signal goes from speaker, to the organic system and finally to the microphone. In this way, the app subtract the two acoustic signals recorded and keep only the organic measurement, that will be registered, saved and transmitted. Finally, the successive measurements will create an historical record of the organic behaviour of the body to get more accurate data.
Fig. 1. TuneVid interface flowchart
3. Priority Statistic Metric (PSM): A valuation system has been created correlating 4 parameters: Location, Environmental Conditions, Air quality, Comorbidity and VI-AID. The location parameter identifies the vulnerability of different zones that could represent hotspots of virus spread. Environmental conditions parameters such as temperature, air pressure , air quality (surface-level ozone, nitrogen dioxide and pollutant particles) are correlated through NASA and other space agencies satellites. Disturbance factors are corrected and linked to the impedance acoustic measurements. The Comorbidity makes reference to the medical history and finally Viral Infections Acoustic Impedance Detection (VI-AID) tool represent the organic measurements.
PSM : Location (10%) + Environmental conditions (10%) + Comorbidity (30%) + VI-AID (50%)
4. Accurate Geolocation: In this final stage the data could be shared in order to mapping and track the spread of the virus which can be used by government organizations to detecting possible hotspots of disease spread.
https://www.youtube.com/watch?v=YhQgThUCwi4&feature=youtu.be
Based on the background discussed above, the project has been divided into two evolutionary stages to facilitate the development of the App. The first stage corresponds to the development of an App with minimal functionality that obtains valuable information to process the curves obtained between various test individuals, and then be able to better train the artificial intelligence network in the next step. The second stage corresponds to the final integration of all the different modules of the App and the interconnections with the server that will carry out the processing. These stages are detailed below.
The first stage of development corresponds to the development of an App with minimal functionalities to be able to obtain valuable information to process the curves obtained between various test individuals, and then to be able to train the artificial intelligence network. In turn, the measurement method was improved by also modifying the measurement scheme to bring the App to a solution that uses a single cell phone where the signal can be generated with a frequency ramp, and which is recorded by the microphone of the cell phone itself. In this way we have the typical configuration to integrate the entire solution into a single application for cell phones. Under this new hardware architecture, the measurement method can be improved by changing to a differential configuration that allows removing and eliminating the distortions introduced by the cell phone hardware to recover only the traveling signal that passed through the ear, nose and mouth. The difference measurement scheme can be seen in the following figure.
Fig.2Differential measurement method.
The differential measurement sequence consists of first performing a calibration measurement (signal B), where a continuous signal of variable frequency is obtained in the range of 1Hz to 22Khz, thereby achieving the transfer function of the Talking set- Microphone. The second measurement corresponds to (signal A) corresponding to the measurement of the speaker-Body (Ear-Nose-Mouth) -Microphone set, in the same continuous frequency range in the range of 1Hz to 22Khz. Finally, the difference between signal A and Signal B is performed, resulting in signal C. This signal (C) corresponds to the Difference Signal, which is the result of measuring signal (A) and correcting it with the signal of Calibration (B) as seen in the following figure.
Fig.3. Result of preliminary test.
In the previous figure you can see the results of the decomposition of the difference in magnitude and phase signals for two different cases of nasal congestion. In the first column, we can see the modulus of the differences in amplitudes between the signals measured in a study subject who had no nasal congestion (first row), and no nasal congestion (second) row. There it can be seen that in the difference signals, the curves are located on the side of the high frequencies (10khz -22khz), while those of low frequency do not influence the measurement. On the other hand, when they decompose the phase of the measurements between nasal congestion and no nasal congestion, it can be seen in the previous graph (column 2) as in the first case, the phase of the signal difference corresponding to the cases without congestion, is located close to the axis of 0 degrees phase (row 1), while in the second case corresponding to the case of nasal congestion, the phase increases continuously and moves away from the phase of 0 degrees (row 2).
From the observations of the previous curves, the potential of this type of differential measurement can be observed because it allows me to evaluate the behaviors of various clinical cases away from low frequencies to locate the largest differences within the high frequency band and ultrasound. On the other hand, the measurement of the phase component allows evaluating the phase shift shown in a very significant form of graph that allows easy discrimination between two similar clinical cases. This conclusion would allow us to approach a new way of performing the measurement, away from the audio frequencies which would allow us to mount our signal on top of another musical signal and thus be able to extract the information from the difference signals.
As explained above after performing the calibration measurement and the measurement of the ear-nose-mouth system, this procedure will allow us to obtain a history of the individual's signals, where after performing the differential measurements they will allow us to enter them into a artificial intelligence algorithm to achieve a more robust identification presumption between cases compatible with Covid19 and other cases. The scheme is concluded with the corrections factor provided by environmental measures of NASA, comorbidity and the geolocation of the mobile phone.
Among the various technologies available to develop mobile applications, TuneVid has been developed using the Flutter environment, which consists of a Google user interface toolkit that allows creating natively compiled applications for mobile, web and desktop devices from a single code base. In turn, Flutter has a unique mobile SDK framework that provides reactive styles without using any Javascript bridge and is ideal for the type of App we are developing, due to the need to geo-locate the application. This SDK allows us to obtain the spatial coordinates of the individual and then proceed to apply the environmental corrections provided by the NASA data. To store all the information required and subsequently processed by TuneVId, it was used as the FireBase database engine, since it itself allows us to access said data in real time. The technical documentation for Flutter and FireBase can be found at Link1, Link2.
Fig.4. Technology App Image.
We decided to use Dart because it is a very flexible programming language where you can write the code and then run it anywhere without any limitations whatsoever.
The Dart code can be executed in four different ways:
The code can be trans-compiled into JavaScript using source-to-source compilation. This is done with the help of the dart2js compiler. As it can be converted to JavaScript, all major browsers support it.
The Dart SDK includes a virtual machine called Dart VM. This mode is for standalone execution. The Dart code can be executed in command line interface mode. The Dart SDK includes a powerful package manager called ‘pub’.
The Dart code can be executed in another mode through the Dartium browser. This is a customised Chromium Web browser that includes the Dart VM. As this browser has direct support for Dart code, there is no need to convert it into JavaScript.
Dart can also be executed in AOT mode. AOT stands for Ahead-Of-Time compilation. In this mode, the Dart code can be directly converted into native machine code.
Also due to the need to geo-locate the application with the spatial coordinates of the individual,be used Here will to enhance these capabilities and integrate it into the final application. Finally, Google AI will be used as an artificial intelligence engine, which has a friendly interface to be integrated into our App. All these tools will allow us to process the information collected by TuneVid and thus be able to apply the environmental corrections provided by the NASA data.
As explained above, various software tools have been developed to complete the different parts of the project.
As for the preliminary tests and the curves measured with the standard applications, these have been processed using a Matlab code found in the following Link.
The code developed during the SpaceApp related to the TuneVid application can be downloaded from the following Link.
while for the information collected from the Health Service of the Government of the City of Buenos Aires, it has been compiled in an Excel file format to be later exported. to the database systems. The proposed format can be consulted at the following Link.
Finally, the information collected from the satellite services has been compiled in an Excel file format to be later exported to the database systems. The proposed format can be consulted at the following Link.
Visit: 31/May/2020
Visit: 31/May/2020
Visit: 31/May/2020
https://gisanddata.maps.arcgis.com/apps/opsdashboard/index.html#/bda7594740fd40299423467b48e9ecf6