iiinova — where your digital self keeps evolving

Tara Eicher @tarabyte1024.bsky.social · 14/01/2026

The Harvard Gazette published an article referencing my work. This piece describes the role that changing diagnostic practices play in the increased rates of autism diagnosis, as well as the importance of having a unified diagnosis under the DSM-5. #autism #research news.harvard.edu/gazette/stor...

news.harvard.edu

Taking a fresh look at definition of autism — Harvard Gazette

Some families, activists say term is too broad, masks unique issues of most severe cases as surging rates, federal plans turn spotlight on disorder

000

Reposted by Tara Eicher

Autistic Self Advocacy Network (ASAN) @autisticadvocacy.org · 05/12/2025

At the recent ACIP meeting, members spread lies about vaccines and autistic people. Vaccines are important. They protect us from diseases, and they let autistic kids grow to become autistic adults. We do not need to be prevented; we need support and respect. autisticadvocacy.org/2025/12/asan...

Orange rectangle over a blue background. Text reads, "For the past two days, the Advisory Council on Immunization Practices (ACIP) met to talk about vaccines and things related to vaccination. At the meeting, ACIP members spread lies about vaccines.

Vaccines are important for protecting the public from dangerous infectious diseases. Vaccines do not cause autism. Vaccines let autistic children live long enough to become autistic adults.

Autism is not a tragedy to be prevented; it is a way of being deserving of support and respect. ASAN will continue to call out and unequivocally condemn any claims to the contrary."

04124

Tara Eicher @tarabyte1024.bsky.social · 06/12/2025

autism at the trait level to develop effective interventions for helping people meet their goals. For example, treating self-injury, cognitive delay, speech delay, etc. rather than “autism” as a whole. This addresses disability directly without pathologizing autistic identity.

110

Tara Eicher @tarabyte1024.bsky.social · 06/12/2025

even those who have low support needs still have some support needs and are therefore still disabled by autism to some extent. Some don’t like to think of themselves as disabled, but that doesn’t mean support is not needed. Instead of creating this dichotomy, we should instead strive to understand

200

Tara Eicher @tarabyte1024.bsky.social · 06/12/2025

Interesting question, Silvaro! I’m actually going to push back a bit here - I don’t think we should create a dichotomy between people to whom neurodiversity applies and those who are “truly disabled” by it. Neurodiversity should apply to even those who are profoundly disabled by their autism. Also,

210

Tara Eicher @tarabyte1024.bsky.social · 18/10/2025

On November 14th at 12 pm ET, I will be giving a talk at the #NHGRI ELSI Friday Forum focused on what #autism #researchers owe the communities they study. I encourage anyone interested in my work, as well as the topics of #neurodiversity or autism research, to register: elsihub.org/ELSIFridayFo...

elsihub.org

ELSI Friday Forum | ELSIhub

000

Tara Eicher @tarabyte1024.bsky.social · 23/09/2025

They do not destroy families. They are not a “horror show” or an “epidemic”. Autistic children *are* beautiful, healthy, and vibrant, just as they are.

100

Tara Eicher @tarabyte1024.bsky.social · 23/09/2025

5. His rhetoric surrounding autism rewinds #advocacy efforts by at least 40 years, to a time when the appropriate response to an autism diagnosis was for a parent to grieve their “lost” child. Autistic children are not lost versions of a “beautiful, healthy, vibrant child”.

100

Tara Eicher @tarabyte1024.bsky.social · 23/09/2025

4. The increase in diagnostic rates is largely due to a better understanding of autism, as well as the removal of speech delay and intellectual disability as diagnostic criteria over multiple iterations of the #DSM.

100

Tara Eicher @tarabyte1024.bsky.social · 23/09/2025

3. Research into the #genetics underlying autism has not fallen short at all. In fact, this research has helped to illuminate the vast complexity of autism genetics.

110

Tara Eicher @tarabyte1024.bsky.social · 23/09/2025

2. #Vaccines do not cause autism, they do not contain #mercury, and there is no evidence that “spacing them out” is safer. There is evidence that it is more dangerous to do so because children go unprotected.

100

Tara Eicher @tarabyte1024.bsky.social · 23/09/2025

1. There is no causal link established between #Tylenol and #autism. Discouraging #pregnant women from taking it and from giving it to babies will result in untreated fevers, which (unlike Tylenol) can cause issues with neurodevelopment.

100

Tara Eicher @tarabyte1024.bsky.social · 23/09/2025

The announcement made today by President #Trump included many scientific errors, as well as dehumanizing and degrading language towards #autistic #children. Here are a few of the problems with his statements and the effects they will likely have:

120

Tara Eicher @tarabyte1024.bsky.social · 20/03/2025

I just need to take a moment to appreciate the fact that this image was in a Nature Research Highlights article. 😆🍔

000

Reposted by Tara Eicher

Carolyn Bertozzi @carolynbertozzi.bskyverified.social · 06/02/2025

Check it out! Another novel RNA modification - ADP-ribosylation - that was previously only known on proteins. A cousin to #glycoRNA 👏 www.nature.com/articles/s41...

Graphic showing modification of mRNA with ADP-ribose mediated by bacterial enzyme cmdTAC

322481

Tara Eicher @tarabyte1024.bsky.social · 07/02/2025

I recently read a paper claiming to address activists’ concerns about ABA, which concluded that the concerns were invalid. A quick Google search of the authors revealed they were all ABA therapists. No COI was declared.

010

Tara Eicher @tarabyte1024.bsky.social · 03/02/2025

I’m excited to see a study that focuses on the impact of interventions on #mentalHealth in #autism. Classically, many studies have focused more on #behavioralOutcomes, something #autismAdvocacy groups have pinpointed as a problem. This study shifts the focus from presentation to internal experience.

000

Reposted by Tara Eicher

Ensembl @ensembl.org · 27/01/2025

Ensembl is 25! 🥳 Join us in celebrating our silver birthday by trying out our new browser - beta.ensembl.org 🎂 Silver carp 🩶🐟 and silvery gibbon 🩶🦧 genomes await!

26321

Tara Eicher @tarabyte1024.bsky.social · 01/02/2025

This is why we need custom #algorithms that account for #biologicalInteractions in an interpretable way.

011

Tara Eicher @tarabyte1024.bsky.social · 01/02/2025

#DeepLearning methods often don’t work well for #genomics research because they are difficult to interpret and require more training data than anyone in the medical field could possibly collect. Classical #MachineLearning, in turn, does not adequately model the complexity of the data.🪡

131

Tara Eicher @tarabyte1024.bsky.social · 01/02/2025

This is why custom algorithms that account for the complex interactions inherent to biology are needed in this space.

000

Tara Eicher @tarabyte1024.bsky.social · 20/01/2025

(1) Pre-trained models (which are necessary for omics data and other small-sample datasets) do not always generalize well to new domains. (2) LLMs do not offer the level of interpretability expected by users in the bioinformatics field.

arxiv.org

010

Tara Eicher @tarabyte1024.bsky.social · 20/01/2025

In short, #LLMs have been applied to biomedical text analysis as well as analytical tasks involving #sequencing data. However, the following challenges still remain: arxiv.org/pdf/2501.06271

arxiv.org

111

Reposted by Tara Eicher

Christie Wilcox @nerdychristie.bsky.social · 17/01/2025

Small AI-designed proteins neutralized one of cobras' nastiest venom toxins. The findings raise the question: Could AI bring antivenoms into the 21st century? That and more of the best from @science.org and science in this edition of #ScienceAdviser: www.science.org/content/arti... 🧪

So-called binder proteins (right), designed by AI, protected cells and mice from lethal doses of 3-finger toxins better than the antibodies used in conventional antivenom (left). Timothy Jenkins

26515

Tara Eicher @tarabyte1024.bsky.social · 11/01/2025

I come across the #fourierTransform in many of the algorithms I study, but it was not intuitive to me when I first encountered it. This video makes it intuitive. #math #engineering youtu.be/r18Gi8lSkfM?...

youtu.be

Fourier Transform, Fourier Series, and frequency spectrum

YouTube video by Physics Videos by Eugene Khutoryansky

000

Tara Eicher @tarabyte1024.bsky.social · 23/12/2024

A nice overview of the problems with Andrew Wakefield’s “research” with a deep-dive into one of his fabricated results. #Vaccines don’t cause #autism. www.youtube.com/watch?v=EXtA...

youtube.com

Wakefield's Smoking Gun

YouTube video by C0nc0rdance

000

Tara Eicher @tarabyte1024.bsky.social · 23/12/2024

#machineLearning research should focus more on solutions for high-dimensional low-sample-size data. The majority of real-world data sets do not have enough samples to train deep neural networks.

000

Tara Eicher @tarabyte1024.bsky.social · 20/12/2024

are “turned off”. In summary, it isn’t our genes that differ (at least not in terms of presence/absence), but rather the ways in which they are accessed and used.

000

Tara Eicher @tarabyte1024.bsky.social · 20/12/2024

The combination of mutations and epigenetic differences accounts for much of the phenotypic difference between individuals, but also variation between cells. For the most part, cells are able to perform distinctive tasks because the genes coding for proteins unnecessary for that cell type 🪡

100

Tara Eicher @tarabyte1024.bsky.social · 20/12/2024

reach the DNA depends on the 3D shape of the DNA (yes, it’s more complicated than a long double-helix). There is an entire field called epigenetics dedicated to studying the accessibility of DNA. 🪡

100

Tara Eicher @tarabyte1024.bsky.social · 20/12/2024

…transcription into RNA is a process that involves several different types of proteins (transcription factors), all of which both need to be present and need to be able to reach the DNA in order for the RNA to be transcribed and the protein to be produced. The ability of transcription factors to 🪡

100

Tara Eicher @tarabyte1024.bsky.social · 20/12/2024

Genes have specific sequences, but those sequences can be modified by mutations, which lead to different proteins being produced, sometimes with similar and sometimes with very different effects when compared to the original protein. But also… 🪡

100

Tara Eicher @tarabyte1024.bsky.social · 20/12/2024

to produce a phenotype. We all *have* all of the genes mostly, but we don’t all produce the same proteins at the same rates. Here’s why: 🪡

100

Tara Eicher @tarabyte1024.bsky.social · 20/12/2024

A little bit of info about #genetics : We often speak of people “having a gene”, but in reality that isn’t how it works. A gene is a location within the DNA that gets transcribed into RNA and then gets translated to make a protein, which then interacts with other proteins and compounds 🪡

100

Tara Eicher @tarabyte1024.bsky.social · 20/12/2024

to produce a phenotype. We all *have* all of the genes mostly, but we don’t all produce the same proteins at the same rates. Here’s why: 🪡

000

Tara Eicher @tarabyte1024.bsky.social · 12/12/2024

To drive network science forward, we need to generalize methods designed for interpreting omics data: hypothesis testing for edge or subnetwork significance, pathway analysis from network input, feature extraction from network input, and sample label prediction from network input.

000

Tara Eicher @tarabyte1024.bsky.social · 12/12/2024

There are a paucity of methods available to interpret them. For people who don’t specialize in network science (most biologists), the result is that you end up with a nice network and don’t know what it tells you. To drive network science forward, we need to 🪡

100

Tara Eicher @tarabyte1024.bsky.social · 12/12/2024

Biological networks provide information about biomolecular interactions that cannot be obtained by analyzing omics measurements (e.g. differential expression) alone. Here is my take on why they are not used more frequently in analysis: NetworkScience 🪡

130