SensiML’s Response to the Coronavirus Pandemic

SensiML's COVID-19 Response

Blog Highlights:

We are living in truly unprecedented times and as COVID-19 impacts us in ways big and small, we are all coping with and adapting to ‘new normals’ in our personal lives and how we do business amidst a worldwide pandemic.

Working from home during the Coronavirus stay at home mandates.
A typical work-from-home office during the stay-at-home pandemic orders

At SensiML, our work building industry-leading AutoML development tools for the IoT edge carries on largely unabated, although like most we are learning to adjust to working from home and all of its challenges.  Thoughts of intelligent auto-muting conference headsets that respond to the acoustics of barking pets and crying babies have come to mind in more than one meeting I’ve attended!

Those minor inconveniences aside, we realize that getting through this will require the effort, courage, and ingenuity of everyone.  Our hearts and deep-felt thanks go out to all of the essential workers – from the doctors, nurses, and first responders on the front line, to the grocery store, logistics, food supply, and other key service workers sustaining our critical needs.

To contribute in the ways SensiML can help, we are announcing two COVID-19 initiatives today.  First, we recognize many of our customers face delays in day-to-day business activities like gaining project and purchasing approvals now made more difficult. So we are making access to SensiML Starter Edition simple with a limited-time $99 online purchase promotion.  SensiML Starter Edition is a fully functional version of the SensiML tool suite at a low introductory price that provides the means for developers to evaluate edge AI algorithms during proof-of-concept projects using their own datasets.  Regularly $499, SensiML is making Starter Edition available for a limited period at just $99 with coupon code “AIFROMHOME” to help developers and engineers quickly evaluate the toolkit using a streamlined online purchase mechanism.  To learn more, see Starter Edition COVID-19 Promo.

Second, we are sponsoring a data collection effort for a pre-diagnostic COVID-19 screening application that uses analysis of coughing sounds to detect probable positive cases.  Currently, most solutions for non-diagnostic coronavirus screening are based on performing temperature scans of individuals.  While fever is certainly a known symptom for positive cases, temperature tests alone are not definitive as many conditions can lead to fever other than COVID-19 and not all who test positive exhibit fever.  Recently published academic research suggests AI analysis of cough noises can be used as an additional means for screening those likely stricken with COVID-19.  As with temperature scanning, cough screening is not a substitute for clinical-grade testing and is not intended as a medical diagnostic.  But, as a decision support tool combined with other indicators  (like temperature scans), cough analysis promises to provide a clearer picture of suspected cases.  For those tasked with administering limited clinical tests, admitting workers into facilities for safe back-to-work efforts, and containing the spread of the virus in healthcare facilities, stores, and other public venues this can be a vital tool in keeping us all safer.

Fever and Cough both addressable leading symptoms for smart IoT screening devices

Existing cloud based AI studies have been demonstrated as viable but come with the limitations of cloud processing detailed in a recent whitepaper we’ve published.  Raw sound data collected locally and submitted for analysis in the cloud suffers from network latency, availability, bandwidth, and privacy factors.  Edge processing of cough data analysis in IoT sensor devices and smartphones as SensiML enables, can provide much quicker classification insight and make standalone devices that can work with or without network connectivity.

Regardless of the processing methods used, more development and data is needed to advance the early academic studies into deployable devices.  To this end, we are getting involved by collecting and publishing an open-source COVID-19 coughing dataset to augment what limited sets have been openly published to date.  You can help us in this effort by contributing your own data at  We need as many samples as possible from people regardless of whether they are perfectly healthy, have other respiratory conditions, or are suspected or confirmed as having COVID-19.  We thank you in advance for contributed data towards this project as well as for spreading the word to your family, friends, and colleagues to contribute as well.

I hope all of you stay healthy, safe, and find new ways to stay productive in your businesses in this challenging time.

Best Regards,
Chris Rogers
SensiML Corporation