How well can models actually interpret clinical notes? The SNOMED benchmark puts them to the test with a large, annotated dataset of real doctor's notes. See how models compare, or submit your own: www.drivendata.org/benchmarks/3...

SNOMED CT Entity Linking Benchmark

A benchmark for linking text in medical notes to entities in SNOMED Clinical Taxonomy.

drivendata.org

Faults are notoriously hard to map, and many across the U.S. remain undiscovered. In the GEMS Prize, solvers are building models to find them using geophysical and elevation data, with a shot at $300K. Deadline: December 3. gems.drivendata.org #GEMSprize

The Geologic Enhanced Mapping System (GEMS) Prize Challenge

Identify hidden geologic faults to discover new geothermal resources.

gems.drivendata.org

All three tracks are still open for the Lost in Transcription Challenge! Compete for $20k in prizes, and put your models to the test transcribing bilingual speech across English-Spanish, Spanish-Nahuatl, and Indonesian-Javanese. Join → competitions.mozilladatacollective.com/competitions...

Clinical NLP gets tricky fast, especially with rare concepts. This benchmark tests how well models can turn clinical notes into structured SNOMED CT data using the largest public dataset of labeled clinical notes out there. Take a look: www.drivendata.org/benchmarks/3...

SNOMED CT Entity Linking Benchmark

A benchmark for linking text in medical notes to entities in SNOMED Clinical Taxonomy.

drivendata.org

The Spanish-English track of Lost in Transcription is open! Transcribe bilingual speech for a shot at $6,000, or $20,000 if you take all three tracks. Join --> competitions.mozilladatacollective.com/competitions...

There's one week left to enter the DaT Parkinson's Challenge! Develop models that predict Parkinson's disease from DaT scans and compete for €25,000 in prizes. Join --> www.drivendata.org/competitions...

DaT Parkinson's Challenge

Help support early and accurate detection of parkinsonian syndromes by developing models that classify dopamine transporter (DaT) scans as normal or abnormal.

drivendata.org

Submissions are open for the Spanish-Nahuatl track of Lost in Transcription! Put your models to the test transcribing bilingual Spanish-Nahuatl speech. $6,000 in prizes here, $20,000 if you take on all three tracks. Sign up to compete → competitions.mozilladatacollective.com/competitions...

Submissions are open for the Indonesian-Javanese track of Lost in Transcription! Put your models to the test transcribing Indonesian-Javanese bilingual speech. Sign up to compete and get notified when the next two language tracks open. Join --> competitions.mozilladatacollective.com/competitions...

One week left to enter Trace the Ace! Help advance how we understand and improve tutoring, with $50K in prizes. Submissions close August 27. platform.k12-ai-infrastructure.org/competitions...

Competition - Trace the Ace

Predict the learning gains from a tutoring session measured by quiz performance in this tutoring outcomes prediction challenge

platform.k12-ai-infrastructure.org

DaT imaging helps diagnose Parkinsonian syndromes, but expert interpretation isn't available everywhere. Build computer vision models that can help, with a shot at €25K in prizes. www.drivendata.org/competitions... @healthdatahub-pds.bsky.social

DaT Parkinson's Challenge

Help support early and accurate detection of parkinsonian syndromes by developing models that classify dopamine transporter (DaT) scans as normal or abnormal.

drivendata.org

A lot of healthcare data is stuck in free text. The SNOMED benchmark focuses on turning clinical notes into structured concepts. Explore it or try your own model: www.drivendata.org/benchmarks/3...

SNOMED CT Entity Linking Benchmark

A benchmark for linking text in medical notes to entities in SNOMED Clinical Taxonomy.

drivendata.org

New challenge! 🔊 Lost in Transcription asks you to build ASR models that transcribe bilingual speech across three language pairs. Sign up now to explore the data, get notified when submissions open, and compete for $20k. Join → competitions.mozilladatacollective.com/competitions...

Trace the Ace is still live!  Build models that predict student learning from tutoring transcripts, and compete for a share of $50K. About 3 weeks left — submissions close August 27.  platform.k12-ai-infrastructure.org/competitions...

Competition - Trace the Ace

Predict the learning gains from a tutoring session measured by quiz performance in this tutoring outcomes prediction challenge

platform.k12-ai-infrastructure.org

"Beautiful" pipelines don't happen by accident. They take intentional structure, cleanup, and iteration over time. Check out our blog post on common pipeline challenges and how to fix them: drivendata.co/blog/pipelin...

5 Challenges of Creating Beautiful Data Pipelines

A look into the hidden complexity of data pipelines, and some suggestions to improve the process.

drivendata.co

We've got a new challenge! Help build models that classify DaT SPECT scans to support Parkinson's diagnosis. €25K in prizes, built on an open dataset from 10 French hospitals, with winning models going open-source. Join → www.drivendata.org/competitions... @healthdatahub-pds.bsky.social

DaT Parkinson's Challenge

Help support early and accurate detection of parkinsonian syndromes by developing models that classify dopamine transporter (DaT) scans as normal or abnormal.

drivendata.org

Structuring clinical notes is one of healthcare AI's toughest problems. See how your model compares in the SNOMED Clinical Entity Linking Benchmark: www.drivendata.org/benchmarks/3...

SNOMED CT Entity Linking Benchmark

A benchmark for linking text in medical notes to entities in SNOMED Clinical Taxonomy.

drivendata.org

Calling all solvers! Join the Trace the Ace challenge and help build models  that predict student learning based on tutoring session transcripts, and compete for a share of the $50k prize pool. Submissions are open now through August 27. platform.k12-ai-infrastructure.org/competitions...

Competition - Trace the Ace

Predict the learning gains from a tutoring session measured by quiz performance in this tutoring outcomes prediction challenge

platform.k12-ai-infrastructure.org

How do you study conversation at scale? DrivenData and BetterUp Labs built the infrastructure to do it, resulting in the CANDOR Corpus - a 1,000+ hour dataset of real-world American English conversations. Read more: drivendata.co/case-studies/building-research-infrastructure-for-conversational-ai

Building research infrastructure for conversational AI

DrivenData partnered with BetterUp Labs to design and build the data infrastructure, machine learning pipelines, and research tools behind the CANDOR Corpus—enabling cutting-edge insights into human…

drivendata.co

Welcome to the K-12 AI Infrastructure Program’s platform - the hub for researchers and developers to access curated K-12 datasets, models, and benchmarks aligned with scientific principles of teaching and learning, and free for everyone to use. Check it out: platform.k12-ai-infrastructure.org

A lot of “data usability” comes down to pretty unglamorous work: cleaning, standardizing, reconciling edge cases - but that’s the part that makes everything else possible. We wrote our approach here: drivendata.co/blog/last-mi...

Solving the last-mile public data problem

Using "baked" data to transform public data repositories into analysis-ready resources

drivendata.co

Results are in for the On Top of Pasketti: Children's Speech Recognition Challenge! 828+ participants tackled kids' voices - one of the hardest problems in speech recognition - and winners cut error rates in half. Meet the winners: drivendata.co/blog/on-top-...

Meet the winners of the On Top of Pasketti: Children's Speech Recognition Challenge

Learn how competition winners, working with one of the largest labeled children's speech datasets assembled, cut transcription error rates in half.

drivendata.co

The $26M K-12 AI Infrastructure Program is launching our new platform! We're issuing grants and hosting prize challenges to create open benchmarks, datasets and models that advance the impact and reach of AI in K-12 education. Explore the platform: platform.k12-ai-infrastructure.org

We added a benchmark focused on clinical entity linking. It uses a large set of de-identified notes annotated by medical professionals, which makes it a pretty realistic testbed. Check out how different approaches compare: www.drivendata.org/benchmarks/3...

SNOMED CT Entity Linking Benchmark

A benchmark for linking text in medical notes to entities in SNOMED Clinical Taxonomy.

drivendata.org

How well can your model understand a doctor’s note? Our benchmark tests how accurately models can identify and link medical concepts using SNOMED CT on the largest public dataset of its kind. Get started: www.drivendata.org/benchmarks/3...

SNOMED CT Entity Linking Benchmark

A benchmark for linking text in medical notes to entities in SNOMED Clinical Taxonomy.

drivendata.org

Excited to be speaking at Good Tech Summit in DC April 7 www.goodtechtogether.org/summit We’ll share a program focused on K-12 education and talk about investing in the foundations of AI: data, models, and benchmarks. We'll explore how these shape AI development in a field. Join us!

The performance gap in children’s ASR is real — and solvable! With 1 week until the April 6 deadline, we’re inviting the global ML community to help close it. Compete for $120K and contribute to better speech technology for kids. kidsasr.drivendata.org

On Top of Pasketti: Children’s Speech Recognition Challenge

Develop cutting-edge ASR algorithms specifically for children's speech to advance early education assessments and teaching tools.

kidsasr.drivendata.org

Children’s speech remains one of ASR’s toughest challenges — and the leaderboard is moving! 3 weeks left to compete for $120K in the On Top of Pasketti Challenge. Submit your model: kidsasr.drivendata.org

On Top of Pasketti: Children’s Speech Recognition Challenge

Develop cutting-edge ASR algorithms specifically for children's speech to advance early education assessments and teaching tools.

kidsasr.drivendata.org

10 years ago, "data science for social good" was just an idea. Today, it's a global movement. Our 10-Year Impact Report reflects on a decade of responsible, real-world AI. Take a look back with us: s3.amazonaws.com/drivendata-p...

Competing in the On Top of Pasketti Word Track? We've published a reference tutorial walking through how to build a model for children's speech recognition — covering data exploration, model training, and submission packaging. Get started: drivendata.co/blog/child-a...

Improving Automatic Speech Recognition for Kids - On Top of Pasketti Word-Track Benchmark

Learn how to train a model to transcribe child speech for the On Top of Pasketti Challenge (Word Track)

drivendata.co

From Bogotá to Almaty, our community continues to impress! IGCPHARMA built prize-winning early dementia prediction models in PREPARE. Kirill Brodt has 23 competitions, 8 top-10s, and 6 top-3 finishes. Two new Community Spotlights: drivendata.co/blog/communi... drivendata.co/blog/communi...