|
Marcel Haas
@harcel.bsky.social
[ View ] |
0 replies 0 reposts 0 likes
|
Marcel Haas@harcel.bsky.social |
308 followers 397 following 317 posts
Data scientist: asst prof in population health @ Leiden University Medical Center. Former astrophysicist: extronomer! Beer sommelier. Pythonista. Aspiring Bayesian. Dad.
|
Marcel Haas
@harcel.bsky.social
[ View ] |
0 replies 0 reposts 0 likes
|
Marcel Haas
@harcel.bsky.social
[ View ] |
Niet alleen arbeidsmigranten, trouwens.
0 replies 0 reposts 2 likes
|
Marcel Haas
@harcel.bsky.social
[ View ] |
Spiders seem to produce quite the offspring!
0 replies 0 reposts 1 likes
|
Marcel Haas
@harcel.bsky.social
[ View ] |
Deze lijkt van 538 te komen, wat een behoorlijk betrouwbare bron is, want een zinnig gewogen combinatie van vele polls: projects.fivethirtyeight.com/polls/presid...
1 replies 0 reposts 0 likes
|
Marcel Haas
@harcel.bsky.social
[ View ] |
I'll be teaching a week of AI and technology to ~350 first year medical students, soon. Very soon, apparently, as my inbox is slowly filling with student emails. 2,5 more weeks and I will get some time back again.
0 replies 0 reposts 1 likes
|
Marcel Haas
@harcel.bsky.social
[ View ] |
"Uitgesproken". Ja, behoorlijk uitgesproken. Ik hoop dat de Kamer er nog niet over is uitgesproken,
0 replies 0 reposts 1 likes
|
Nadine Böke
@nadineboke.bsky.social
[ View ] |
Echt, what could go wrong? ☹️
nos.nl/artikel/2524...
1 replies 1 reposts 5 likes
|
Marcel Haas
@harcel.bsky.social
[ View ] |
What AFAS did is nice for their employees, no doubt about that. That they can do it because they charge their customers more than 20% more than necessary shows to me that they're not quite the philanthropists they now claim to be.
0 replies 0 reposts 2 likes
|
Marcel Haas
@harcel.bsky.social
[ View ] |
The new (version >= 01.5) sklearn TunedThresholdClassifierCV looks really great: scikit-learn.org/stable/modul...
It's like a cross-validator for yuning the decision threshold in binary classification based on a standard or custom cost function.
0 replies 0 reposts 0 likes
|
Marcel Haas
@harcel.bsky.social
[ View ] |
If you voted against climate action this weekend and you have children: admit and explain it to them tomorrow. I bet you they will remember this in a few decades from now.
0 replies 2 reposts 4 likes
|
Marcel Haas
@harcel.bsky.social
[ View ] |
Thinking ahead doesn't seem to be the prevailing skill in the majority of the electorate....
1 replies 0 reposts 1 likes
|
Marcel Haas
@harcel.bsky.social
[ View ] |
Green parties seem to lose pretty badly, almost everywhere in Europe. Isn't the planet rather clear about how bad the climate situation is? How does migration always seem to win the race for attention?
1 replies 0 reposts 5 likes
|
Marcel Haas
@harcel.bsky.social
[ View ] |
Dark days for the EU.
0 replies 0 reposts 2 likes
|
Marcel Haas
@harcel.bsky.social
[ View ] |
Oh my, those exit polls don't look great....
0 replies 0 reposts 0 likes
|
Marcel Haas
@harcel.bsky.social
[ View ] |
En bij jou? Wat zijn je zorgen? Wat voor stem werd dat?
0 replies 0 reposts 0 likes
|
Marcel Haas
@harcel.bsky.social
[ View ] |
Ik ben zo'n kiezer en kom uit Leiderdorp (niet dat ik snap waarom dat relevant is). Mijn zorgen? Het voortbestaan van onze kenniseconomie, de staat van het klimaat, de veranderende toon van politici en dito ruwheid in de samenleving, de toenemende ongelijkheid in rijkdom, gezondheid... En meer!
1 replies 0 reposts 5 likes
|
Marcel Haas
@harcel.bsky.social
[ View ] |
Sterker nog: als ik het anders doe, en dat pobeer ik wel eens, dan krijg ik steevast commentaar op mijn niet-academische taalgebruik en artikelopbouw. Men denkt dat editors en reviewers het ook niet anders willen. Vooral de opbouw van een paper jeukt bij mij.
1 replies 0 reposts 6 likes
|
Marcel Haas
@harcel.bsky.social
[ View ] |
PSA: recruiter != consultant
0 replies 0 reposts 0 likes
|
Marcel Haas
@harcel.bsky.social
[ View ] |
You must be kidding me. Due to an "unfair system" (tax paid over expected gain on capital, which may not be the realistic gain), the richest of our country now get a good part of their taxes back, so they pay even less. Billions of Euros back to those who can miss it most. Sadness.
0 replies 0 reposts 0 likes
|
Marcel Haas
@harcel.bsky.social
[ View ] |
I am on the lookout for (affordable) options of creating a (roughly 30 sec) AI generated movie using a textual prompt. Any pointers? Thanks!
0 replies 0 reposts 0 likes
|
Marcel Haas
@harcel.bsky.social
[ View ] |
@groenlinks-pvda.bsky.social Jullie zouden het op je site wel echt een stuk makkelijker kunnen maken om te zien welke kandidaat bij welke onderliggende partij hoort, gezien het belang ervan in de Europese fracties. Help je kiezers!
1 replies 0 reposts 1 likes
|
Marcel Haas
@harcel.bsky.social
[ View ] |
Go vote! Europe needs it, and not voting is equivalent to saying that the large problems we face are irrelevant. Moreover, not voting for a climate active party will make the world of our future generations insufferable.
0 replies 0 reposts 0 likes
|
Marcel Haas
@harcel.bsky.social
[ View ] |
Dan ben je meestal gewoon een beetje laat. Denk ik?
0 replies 0 reposts 2 likes
|
Marcel Haas
@harcel.bsky.social
[ View ] |
Sunshine, please!
0 replies 0 reposts 1 likes
|
Marcel Haas
@harcel.bsky.social
[ View ] |
A model pre-trained on the corpus itself strongly outperforms string matching (the conventional method). A great strategy: translate to English and use an English clinical language model, fine-tuned for the prediction task. Promising results, also for other (perhaps less popular) languages! (N/N)
0 replies 0 reposts 0 likes
|
Marcel Haas
@harcel.bsky.social
[ View ] |
This can be very useful, as the structured data in Electronic Health Records are often very incomplete in these features, so this is a promising way to enhance the completeness and quality of such EHRs. This was achieved by training and fine-tuning BERT models to identify the best strategy. (3/N)
1 replies 0 reposts 0 likes
|
Marcel Haas
@harcel.bsky.social
[ View ] |
Hielke extracted smoking status (current/previous/no user) and correspondingly for alcohol and recreational drugs, from clinical notes of the HagaZiekenhuis hospital. This can be very useful, as the structured data in Electronic Health Records are often very incomplete in these features. (2/N)
1 replies 0 reposts 0 likes
|
Marcel Haas
@harcel.bsky.social
[ View ] |
#PaperDay! 🧪
The first paper of the first PhD student I supervise is out! Hielke Muizelaar studied a practical use case of Natural Language Processing in healthcare. Read it here (open access, including the reviewer reports and responses!).
A thread about the scientific content: (1/N)
1 replies 0 reposts 6 likes
|
Marcel Haas
@harcel.bsky.social
[ View ] |
So let me get this straight. Trump will be allowed to run for president, but in that same election he cannot vote himself, because he is a convicted criminal? So we don't trust him enough to vote, but we do trust him enough to occupy the oval office? The logic escapes me.
0 replies 0 reposts 1 likes
|
Marcel Haas
@harcel.bsky.social
[ View ] |
It's equal parts sad and frightening that NATO feels the need to show muscle, just because that might make it less likely that Russia does (even more) crazy things in the near future (directed at NATO members).
0 replies 0 reposts 0 likes
|
Meg Schwamb
@megschwamb.bsky.social
[ View ] |
Did you ever want to download a AAS Research Note as a PDF? That feature is now available according to @Chrislintott@mastodon.social . yay! "It's hidden under the 'Article and Author information' tab, but it's there! "
0 replies 2 reposts 8 likes
|
Marcel Haas
@harcel.bsky.social
[ View ] |
Apparently, ESA doesn't really like the result.
0 replies 0 reposts 0 likes
|
Marcel Haas
@harcel.bsky.social
[ View ] |
True, but unlike anything else, when a particle moves directly through, it has 0% chance of getting out (which in the case of WIMPS would be very different from anything else of a solar mass, where only a few particles interact). A SMBH has a Schwarzschild radius the size of a solar system...
1 replies 0 reposts 0 likes
|
Marcel Haas
@harcel.bsky.social
[ View ] |
An accretion disk won't even form. But particles (if they're particles) will fly by (slowly, if DM is cold enough) and they can 'hit' the event horizon, right? Is the cross section for that so small that this will make much too small an accretion rate?
1 replies 0 reposts 0 likes
|
Marcel Haas
@harcel.bsky.social
[ View ] |
I don't know if it is!
0 replies 0 reposts 0 likes
|
Marcel Haas
@harcel.bsky.social
[ View ] |
Why can't early black holes not grow by eating dark matter? No radiative pushback, but perhaps too small a cross-section? Would that depend on DM temperature and particle mass, but already excluded by candidate parameters? I thought I was trained in this business, but... 🤷🏼♂️ 🔭
2 replies 0 reposts 0 likes
|
Marcel Haas
@harcel.bsky.social
[ View ] |
I believe a part of the scientific Python stack is still in Fortran?
1 replies 0 reposts 1 likes
|
ELSA: Euclid Legacy Science Advanced analysis tools
@elsa-euclid.bsky.social
[ View ] |
The Euclid consortium have a big spash coming tomorrow... Watch this space for the first results from Thursday, 23 May, 12:00 CEST, and shortly afterwards on arXiv... #EuclidLeaks #science #astronomy #astrodon #space 🔭🧪
1 replies 7 reposts 8 likes
|
Marcel Haas
@harcel.bsky.social
[ View ] |
I just discovered this fun monthly podcast: The astrophysics podcast. Give it a try! 🔭
0 replies 1 reposts 5 likes
|
Marcel Haas
@harcel.bsky.social
[ View ] |
Met 50/50 staan ze ook beide minder sterk, op internationaal niveau...
0 replies 0 reposts 0 likes
|
Marcel Haas
@harcel.bsky.social
[ View ] |
Bij 80/20 zou dat in principe ook nog kunnen...
1 replies 0 reposts 0 likes
|
Lisa Westerveld
@lisawesterveld.bsky.social
[ View ] |
Forse bezuinigingen op: - Onderwijs en wetenschap; - Cultuur; - De vrije pers; - De meest kwetsbare groepen. Maar als je daar kritiek op hebt ben je volgens de napraters een ‘jankerd’. Echt mensen, denk zelf even na over wat hier gebeurt.
21 replies 81 reposts 216 likes
|
Astronomer Royal for Scotland | Catherine Heymans
@astroroyalscot.bsky.social
[ View ] |
At 11am (UK) today, the Royal Astronomical Society releases their long-awaited report on bullying and harassment in the field. It's going to be a hard read with the headline "44% of the 661 survey respondents reported suffering [bullying and/or harassment] in the workplace" 🔭👩🔬 👇Join Online👇
1 replies 12 reposts 20 likes
|
Marcel Haas
@harcel.bsky.social
[ View ] |
Computers will always do exactly what you tell them to do, and if you're lucky, they'll also do what you want them to do. If you're unlucky: try again, but think first. Thinking helps. This is advice for a friend, obviously.
0 replies 0 reposts 1 likes
|
Marcel Haas
@harcel.bsky.social
[ View ] |
Good luck!
0 replies 0 reposts 1 likes
|
Marcel Haas
@harcel.bsky.social
[ View ] |
You forgot the decimal sign though. Rounded off, it's 257 million years.
1 replies 0 reposts 1 likes
|
Marcel Haas
@harcel.bsky.social
[ View ] |
It was the unit conversion.... Right...?
1 replies 0 reposts 1 likes
|
Marcel Haas
@harcel.bsky.social
[ View ] |
I had a blast at the Dutch Astronomers Conference yesterday. Speaking about career paths to Early Career Researchers was fun and it was great to catch up with old friends!
#astro #NAC2024 #DutchAstronomersConference #ECR #Career
marcelhaas.com/index.php/20...
0 replies 0 reposts 5 likes
|
Marcel Haas
@harcel.bsky.social
[ View ] |
Some things are just exactly the same. The level of the talks, including the visuals used, definitely improved over the years. Still the same lovely vibe of like-minded people. It was great to be back for a day.
0 replies 0 reposts 0 likes
|
Marcel Haas
@harcel.bsky.social
[ View ] |
I'll be speaking in an Early Career Researchers session about my career path and choices, which took some twists and turns. Would have been great to have had this option 11-14 years ago, when some of the most significant turns were taken.
0 replies 0 reposts 1 likes