Let's talk about the possibilities of scientific software in routine care and the potential of #OpenSource in #Healthcare!
We often utilize conferences, alongside our own presentations, for privately arranged meetings in small groups, to discuss our #AMPEL #CDSS project. At the end of the month, we're heading to #Rome for the first time!
Feel free to drop by Maria's session or let's schedule a meeting: https://www.uniklinikum-leipzig.de/einrichtungen/ampel/en/contact
How does the #Future of #Medicine look? And how can we help shape it through the #AMPEL project? We had the opportunity to discuss this on the German public broadcaster #MDR. Result: AI will become an integral part of medical practice!
(from min 46:30)
https://overcast.fm/+HCzVKXb3A/46:29
How can we get more #OpenSource into medicine, especially into the medical device landscape? Martin Federbusch (Project Leader of AMPEL) led a discussion at Analytica in Munich on this topic. The occasion was the Open Source release of the #AMPEL #Sepsis Early Detection Model based on the complete blood count. Open Source in routine care is gaining momentum: it's no longer about the "Why," but about the "How"! đź’Ş
Let's do it!
Secret tip for today's analytica in Munich: At 3:00 PM (ICM - Room 4a), the session "Applications of AI Algorithms in Laboratory Medicine (DGKL)" will kick off with Martin Federbusch (Head of AMPEL-CDSS). Surprise - it will focus on AMPEL, but also on many concrete AI applications, such as our recently published AI model for early detection of sepsis, based solely on the complete blood count.
https://veranstaltungen.gdch.de/microsite/index.cfm?l=11543&sp_id=2&selSiteID=vplanner
Spring is coming, and the birds are starting to sing again. Creating a spectrogram from bird chirps and then using an #OpenSource #ImageRecognition to identify the bird - now possible on every smartphone.
https://www.tu-chemnitz.de/tu/pressestelle/aktuell/10606/en
InnoSpeedDating aims to connect young researchers and foster new project ideas.
https://www.leistungszentrum-smart-production-materials.de/start.html
Thank you very much also for the great feedback on our #AMPEL project, it motivates us every time again!
[2/2]
Thank you so much for the fantastic AMPEL workshop at this year's #SPIN2030 science festival! We had a great time 🙂
Interested in exciting scientific projects? We were able to discover some new ones:
While others desperately try to capture plastic in the ocean with large floating rings, an innovation from Leipzig could bring about a turning point: plastic degrading enzymes!
https://research.uni-leipzig.de/bact/technology-transfer-and-cooperation/start-up-projects-n/
[1/2]
We found:
- The diagnostic value of the CBC is significantly higher than assumed
- Even the basic model can achieve predictive power of up to 15% PPV [at 2% sepsis prevalence]
- Procalcitonin, as most established sepsis parameter, can be significantly surpassed when combined with the AI model [AUC: 0.857 vs 0.790]
Article in “Clinical Chemistry”:
https://doi.org/10.1093/clinchem/hvae001
English press release of the University of Leipzig Medical Center:
https://www.uniklinikum-leipzig.de/einrichtungen/ampel/en/Pages/sepsis_press_release.aspx
[2/2]
Delayed sepsis detection increases mortality risk. How can #MedicalAI assist us here?
In the #AMPEL project, we extensively researched this over years using multicenter data from Germany and the USA. The results of our #OpenSource #AI basic model, based only on few parameters of the complete #BloodCount (CBC), exceeded expectations:
[1/2]
#Sepsis #SepsisAwareness #DeutschlandErkenntSepsis #GemeinsamGegenSepsis #GetAheadOfSepsis
🚨 Today we published our #OpenSource AI model for the early detection of sepsis in the world's leading journal of laboratory medicine!🚨
Discovering sepsis patients sooner (=saving lives) is now possible with just a bit of routine lab data
Article:
https://doi.org/10.1093/clinchem/hvae001
Press release of the University Hospital Leipzig [German]:
https://www.uniklinikum-leipzig.de/presse/Seiten/Pressemitteilung_7850.aspx
Press release [English]:
https://www.uniklinikum-leipzig.de/einrichtungen/ampel/en/Pages/sepsis_press_release.aspx
#Sepsis #SepsisAwareness #DeutschlandErkenntSepsis #GemeinsamGegenSepsis #GetAheadOfSepsis
Experience firsthand how AI improves medical care in a clinic?
You can do just that on March 8th at the #SPIN2030 #ScienceFestival in #Dresden in our AMPEL workshop.
With a kind of Pen&Paper game, we'll go through all the pitfalls, and with a bit of luck, we'll successfully integrate AI into the clinic together!
Even outside the workshop, we're happy to introduce the #AMPEL-CDSS!
Contact: ampel@uniklinik-leipzig.de
https://wissenschaftsfestival.lineupr.com/wissenschaftsfestival/schedule
For over 5 years now, AMPEL has been researching automated AI models in the clinical everyday life of the University Hospital Leipzig! #HappyBirthday
We've chosen to forge a new path in #Healthcare, advancing AI in medicine alongside the #OpenSource community
Our #OSS initiative kicks off with the gradual release of all 13 existing #AMPEL models
https://www.uniklinikum-leipzig.de/einrichtungen/ampel/en/models/overview
Our aim is to make the content of our models easily understandable and accessible to all. We truly appreciate your feedback!
One-third of all inpatients experience hypo- or hypercalcemia. This can potentially be overlooked. We have researched how to monitor severe #Calcium imbalances automatically using clinical routine data (see previous post).
We have now published a lay-friendly overview (admittedly, sometimes even our heads start smoking here) on our homepage:
https://www.uniklinikum-leipzig.de/einrichtungen/ampel/en/models/calcium
As part of our Clinical Decision Support System #CDSS, we are currently evaluating the new cutoffs in our clinic and encourage others to do the same.
#AMPEL can support physicians with the most reliable screening and enhance #PatientSafety.
The AMPEL-CDSS is an #OpenScience project, so get involved! And yeah, we also do #ML and #MedicalAI ;)
More about the project: https://www.uniklinikum-leipzig.de/einrichtungen/ampel/en
[2/2]
Using automation in #Healthcare doesn't always require #ML and #MedicalAI. Often, it is sufficient to implement existing knowledge in a more easily applicable form #CDS
We investigated the use of total (tCa) and albumin-adjusted #Calcium (aCa) in severe calcemia detection.
We found:
• aCa is generally superior to tCa [as expected]
• Local hospital cutoffs are crucial [as known but widely ignored]
• We can now prevent >85% of all tCa misclassifications [wow!]
https://doi.org/10.1515/cclm-2023-0805
[1/2]
Labor medicine holds a significant place in Germany. Many #ML models rely on #labdata, and we believe this sets the stage for exciting #AI applications in the future #healthcare system.
The German Laboratory Medicine Congress #DKLM has been taking place since yesterday. Our contribution on #bloodcount-based ML early detection of #sepsis won the 1st prize today. Thanks for valuing our work! 🏆
The awarded poster can be downloaded here: https://www.uniklinikum-leipzig.de/einrichtungen/ampel/Freigegebene%20Dokumente/Sepsis%20Poster%20DKLM%202023.pdf
Using our #AI frameworks for Acute Kidney Injury (#AKI) and #Sepsis as examples, we will present this again on October 12th and 13th at the German Laboratory Medicine Congress in Mannheim. #DKLM
https://laboratoriumsmedizin-kongress.de/
For all those who we cannot reach in this way (or would like to read in advance), we provide our posters as PDFs here:
We appreciate your feedback!
#AMPEL #CDSS #MedicalAI #PatientSafety #CDS
2/2
How can #AI be implemented in #healthcare? We had some great discussions about this at the Saxon Hospital Day in Leipzig and at the German Association of Medical Informatics Meeting (#GMDS) in Heilbronn. Thank you for the valuable input!
We are pleased that our research- and common good-oriented approach #AMPEL has generated such significant interest. Let's achieve this goal together!
1/2
When it could have been avoided, that's when you get the most frustrated. #PatientSafety is precisely that.
A few years ago, the WHO began prioritizing this issue and established #WorldPatientSafetyDay on September 17th.
We are among the projects actively striving to fulfill this mission with #CDSS :)
The #AMPEL team develops a #Sepsis model based on #LabData. It has been shown that just a few parameters from the blood count may suffice: https://doi.org/10.1101/2022.10.21.22281348
Preliminary findings indicate that in specific scenarios (e.g., #ED), even better predictions can be achieved.
Next year, we will release our #OpenSource #CDSS, including an ML model for #Sepsis detection!
2/2