Page - 323 - in Intelligent Environments 2019 - Workshop Proceedings of the 15th International Conference on Intelligent Environments

parameters suchas the inhalationcapacity (IC)and the forced inspiratoryvolume inone second (FIV1), areeasilyderived. ExperimentNo. IFR(L/min) IC(L) FIV1(L) Estimated Actual RMSE Estimated Actual RMSE Estimated Actual RMSE E1 25.424 25 0.4241 2.058 2.0243 0.034 0.0032 0.0042 1e-03 E2 31.258 30 1.258% 3.134 3.008 0.126 0.0032 0.0050 0.001 E3 39.619 40 0.381 3.972 3.837 0.135 0.0034 0.0067 0.003 Table1. Comparisonof theestimated inhalationparameters IFR, IC,FIV1and their actualvalues Table 1 compares the actual values vs. the estimated values of the inhalation pa- rameters. This table also includes the rootmean square error (RMSE) as an important indicator of the accuracy.TheRMSEonFIV1estimationwas slightlyhigher compared to other parameters; this can be explained by the amplitude of the acoustic inhalation signal. Since the energy signal is correlatedwith IFR,we can see in Figure 4 that low amplitudes correspond to the first seconds of inhalation. Therefore, the observed error are due to the fact thatwe assumeda steady test flowwhenestimating the actual FIV1. However, theoverallRMSEforIFRandICisremarkablylow,whichillustrate themerits of theproposedmethod. 4. Conclusionandfuturework Most relatedworks on acousticmonitoring systems requiremicrophones to be placed near tothesoundsource,whichcanbesometimesuncomfortable tousefor tothepatient. In thiswork,wepropose anew technology tomonitor inhalationperformanceusing an acousticdevice.Usinganinhaleradd-onandasmartphone,wewereable tocapturegood quality acoustic signals and accurately estimate different inhalation parameters. These promising results motivate us to incorporate the proposed technology into an acoustic monitoringsystemforasthmapatients, thatwill includeaninhalerwithanadd-ondevice (both DPIs and pMDIs) which in turn is connected to a smartphone via an m-health application.Thiswouldprovideasthmapatientswithall the important instructionswhen using their inhaler in case of a non-adherence, and also help themgain control of their asthma condition and prevent potential exacerbations by allowing them to have a close look at their lung function through the inhaler sensor. In continuation of this research, we are planning to test this new technology on asthma patients in the region of Rabat (Morocco)andwe intend to includeother relevantdata inourpredictionmodels suchas the patient’s self-reports,weather and air qualitymeasurements through ageographical information system (GIS). These clinical trialswill enable us to improve the prediction quality of ourmodels, thusmaking this technology very useful in the field of asthma managementandcontrol. References [1] The Global Asthma Report 2018. (n.d.). Retrieved March 31, 2019, from http://www.globalasthmareport.org/ [2] Global Initiative forAsthma -Global Initiative forAsthma -GINA. (n.d.).RetrievedMarch31, 2019, fromhttps://ginasthma.org/ Z.Jeddi etal. /Estimationof InhalationFlowParameters forAsthmaMonitoring 323