De titel zegt het al, en er staat een mooi bericht online waar ik graag naar verwijs, maar wil persoonlijk graag toevoegen dat ik dit een geweldige mogelijkheid biedt om vanuit Healthplus.ai samen met PW Consulting en SAS aan de slag te gaan om de verdere ontwikkeling van PERISCOPE gestalte te geven!
On cleaning GPS data and Observability For Good
Just a short note on Pinetrail, a very nice tool I used today to test sanitizing GPS log files to correcting the elevation, and remove outliers and idle points. “But GPS is so accurate, do we still need that?” Yes we do, because even nowadays there is still some glitches and quirks due to mobile devices jumping to GSM transmission towers, occasional noisy signals or no signal at all, unknown app errors etc. And cleaning out the idle points also improves the visual representation.
Reason for testing in the first place is Sensing Clues, where the field rangers collect vast amounts of route data with observations and actions when on patrol. This data needs to be interpreted and there it helps to have clean(ed) data. Most probably there are also some additional rules needed but Pinetrail looks like a good starting point, also because it provides a nice report of what it does as you can see in the example above, on one of my running routes I exported from Strava (I really did not start my run with these strange figures).
If you have to clean some dirty GPS data too, head to https://github.com/sosna/pinetrail and give it a try.
Oh, and there is some more interesting news coming in a next post on my work for Sensing Clues. As we try to provide the best technical solutions on the web and mobile for the field rangers in the wildlife parks, this also means we need to be sure everything is up and running the whole time. In order to keep a clear overview on that, we need a proper full stack monitoring tool “from mobile to database”, “from network to Docker container” and everything in between.
So Sensing Clues applied for the New Relic Observability for Good program, and the approval came back almost instantly! A very interesting free program for nonprofits, charities and NGOs (or ANBIs in proper Dutch). Currently installing the first monitoring Agents, more on the results soon.
Revamped website healthplus.ai
And then just before the year ends the website of healthplus.ai is on par with the business ambitions again! For those of you who visited the -now- old website, there is quite some change: clear statements on what we are working on, fresh colors and a nice layout that contribute to the user experience and make it easier to find the info you need.
Behind the screens we are also working hard on the platform itself of course, so keep coming back to the website every now and then for more news.
Sensing Clues Focus 2.0 live!
Afgelopen vrijdag is Sensing Clues Focus 2.0 live gegaan als onderdeel van de migratie naar het nieuwe WITS (Wildlife Insights tool suite) platform.
Een niet geringe inspanning die zowel herontwerp van de frontend web applicatie als upgrade van de backend database inhield.
Proficiat voor het hele Sensing Clues team van medwerkers en vrijwilligers met dit resultaat, en nog belangrijker wat dat voor de wildlife parks gaat betekenen!
Wanted: launching customer! (or: ‘proefkonijn gezocht’)
If I would tell you that we are getting close to release an MVP for “the next generation self service cloud based solution that enables your organisation to fulfil urgent and valuable enterprise data related requirements in an actionable way”, would you be interested?
I bet you are!
So, just grab the phone or send me an email if you want more information.
At this moment I cannot reveal more but after signup for a trial under NDA, you get all the details you need.
Teaser…
Now I’m back on the writing track I might as well just add a first teaser about another project I’m working on. It is not a home alone project, quite a bit more than that, and it includes some serious technological challenges.
Here is the teaser:
Curious to hear what you think of it. Feel free to post your ideas in the comments. There will be healthy rewards for whoever gets close to the actual concept, and also for best contribution.
Healthplus.ai PERISCOPE
It took a bit longer than expected to actually write the next post but whoop, here it is, and there is some nice news!
I started on a freelance engagement assisting Healthplus.ai in translating hospital data into processable data streams and implementing a scalable approach for each and every hospital.
In due time this should result in PERISCOPE, an EHR-integrated clinical tool that uses AI for early warning of post-surgery infections and other major complications. Earlier diagnosis and actions will reduce the number complications, their impact and cost.
Sounds good right? If you want to know more (and there is much more because this project already started some time before I joined the team but more on that in another post), check out https://www.linkedin.com/company/healthplus-ai/ and https://healthplusai.com/
Interesting to mention is that Healthplus.ai is mentioned in the current AI Startup Landscape map, check out the EHEALTH section in the overview of 100 Dutch AI startups.
Waking From Hibernation, the Hard Work Begins
Yes, I more or less stole the above header from a NY Times article (https://www.nytimes.com/2017/03/24/science/hibernation-spring-bears-bees-bats-arctic-squirrels.html) as I needed a tag line to express how coming back from a long dormant period feels. The time since the last post was mainly filled with finishing building the house, and making it livable (last piece of resistance is the garden), lots of the normal work-work, and for the last few months of course the very sad corona-situation that blocked and changed a lot of things too.
Now the time has come to start doing some more data exercises again, preferably geospatial but I won’t refuse to take up other data challenges too, like linked data (wrote a short post on that a while ago at LinkedIn, mmh need to follow up on that one too).
I guess the first thing to come is some data exploration in the coming week. Stay tuned, I promise I won’t fall asleep that long again.
By the way, the NY Times article might be picked less random than it seems. I’ll use it to link to some of the latest work that Sensing Clues has done recently together with Rewilding Europe on a pilot project in the Central Apennines to predict bear behaviour, read all about it at https://rewildingeurope.com/news/pilot-project-in-the-central-apennines-looks-to-predict-bear-behaviour/
Sensing Clues is a non-for-profit organisation working with several IT partners that provide technology and free consultancy to create a wildlife intelligence platform that helps to prevent human-induced threats for nature like poaching and illegal logging, and to improve peaceful human-wildlife coexistence.
Sensing Clues is always looking for more help in the form of technologies or services so if you feel you can help, let me know or contact them directly.
Bathtub Sessions – About to launch 2 apps
Some companies have their Data Lab to develop innovative ideas, I have my bathtub that serves the same purpose. It is a portable concept, because for me it works not only with bathtubs but with showers, swimming pools and saunas too so I can also have one of those sessions at the pool during holidays, or when freshening up in my hotel room on a business trip. Latest project I’m working on, is two dating apps: one for the nobility called “Noblesse Oblige” and one for pets called “Animal Sana”. Both rely heavily on machine learning for automatic classification and verification of genealogy documents, which are important for both the nobles with their family tree, and pets with their pedigree. Of course, this is not only to ensure that the submitted documents are authentic, but more important as input for the matching algorithms (preventing inbreeding is a big thing).
Target audience for “Animal Sana” is cats and dogs first, later on to extend to rodents and horses. For nobility there is global coverage from the start, based on local (national) entitlements of course. Except US unfortunately, since the Constitution of the United States prohibits both the United States and the separate states from granting titles of nobility. I might setup a similar but separate app for the US at a later time, not sure yet.
The apps use a generic underlying framework that includes artificial intelligence for facial recognition to distinct between nobles and pets. It works now with a 98,5% accuracy so there is hardly any chance left that a noble’s profile gets linked to a pet’s profile or vice versa (had some issues discriminating between horses and knights in early versions). I’m now working on finetuning some components before first public versions will be released. Main challenge is to create a responsive UI that works for both nobles and pets. Previous A/B testing showed that pets were more comfortable with a rather complex UI for both mobile and web, and preferred advanced and detailed selection methods in addition to initial AI based match making, while this was rather hard to grasp for the nobles. So further work needs to be done here too to create a user friendly yet powerful interface.
In order to move to massive scale directly from the start, and to “put the dots on the i” as we say in the Netherlands, I’m looking for investors who bring money, time or whisky (in random order) to the table. If you want to be part of this adventure and help me with the finishing touches let me know at bizdev@basaltaura.nl
Makeover https://www.psychologenpraktijkfleuren.nl
Recently did a makeover for https://www.psychologenpraktijkfleuren.nl, so if you are curious have a look…
