We tested the method on history first. Of the fields our 2018 map flagged, 28% broke out by 2024, nearly three times the rate for science as a whole.
The exact rule behind this list, run on the 2018 map, picked sixteen fields. Ten of them broke out.
Published October 2026.
Text-to-image diffusion models
The models behind AI image generators such as Stable Diffusion.
Papers, 2021 to 2024
114 to 2,144
18.8x
Other fields citing it, 2021 to 2024
29 to 598
20.6x
Human-centered manufacturing (Industry 5.0)
Manufacturing research that designs automation around the people working beside it.
Papers, 2021 to 2024
221 to 893
4.0x
Other fields citing it, 2021 to 2024
26 to 212
8.2x
Underground hydrogen storage
Storing hydrogen at grid scale in salt caverns and depleted gas fields.
Papers, 2021 to 2024
154 to 527
3.4x
Other fields citing it, 2021 to 2024
30 to 85
2.8x
Piezocatalysis for pollutant breakdown
Materials that build up charge under vibration or ultrasound and use it to break down pollutants.
Papers, 2021 to 2024
243 to 622
2.6x
Other fields citing it, 2021 to 2024
72 to 190
2.6x
Physics-informed neural networks
Neural networks trained to obey the equations of physics, used to simulate fluid flow and other physical systems.
Papers, 2021 to 2024
626 to 1,837
2.9x
Other fields citing it, 2021 to 2024
340 to 744
2.2x
Color-changing freshness films
Food packaging that changes color as the food spoils, using plant pigments called anthocyanins.
Papers, 2021 to 2024
202 to 577
2.9x
Other fields citing it, 2021 to 2024
95 to 197
2.1x
Green ammonia production
Making ammonia from renewable hydrogen, for fertilizer and as a fuel that stores energy.
Papers, 2021 to 2024
90 to 195
2.2x
Other fields citing it, 2021 to 2024
26 to 63
2.4x
Hydrogen-bonded organic frameworks
Porous crystals held together by hydrogen bonds, built for gas separation and catalysis.
Papers, 2021 to 2024
105 to 228
2.2x
Other fields citing it, 2021 to 2024
52 to 108
2.1x
Mendelian randomization
Using inherited genetic variants as natural experiments to test whether an exposure causes a disease.
Papers, 2021 to 2024
767 to 2,927
3.8x
Other fields citing it, 2021 to 2024
535 to 1,456
2.7x
Plant disease detection from leaf images
Deep learning that spots crop disease in photographs of leaves.
Papers, 2021 to 2024
1,310 to 3,708
2.8x
Other fields citing it, 2021 to 2024
245 to 609
2.5x
Machine-learning autism screening
Models that screen for autism from eye tracking or brain scans.
Papers, 2021 to 2024
232 to 455
2.0x
Other fields citing it, 2021 to 2024
28 to 60
2.1x
Microneedle drug delivery
Patches of tiny needles that deliver drugs through the skin or sample the fluid beneath it.
Papers, 2021 to 2024
518 to 959
1.9x
Other fields citing it, 2021 to 2024
234 to 394
1.7x
cGAS-STING signaling
The pathway cells use to sense stray DNA and raise an immune alarm, now a drug target in cancer and inflammatory disease.
Papers, 2021 to 2024
581 to 1,103
1.9x
Other fields citing it, 2021 to 2024
494 to 776
1.6x
Oxidative balance scores
Composite diet and lifestyle antioxidant scores tested against disease risk in large population surveys.
Papers, 2021 to 2024
58 to 194
3.3x
Other fields citing it, 2021 to 2024
25 to 87
3.5x
AI reading of dental X-rays
Deep learning that finds decay and bone loss in dental radiographs.
Papers, 2021 to 2024
383 to 1,111
2.9x
Other fields citing it, 2021 to 2024
48 to 125
2.6x
Digital health literacy
Whether patients and clinicians can find, judge and use digital health tools.
Papers, 2021 to 2024
264 to 511
1.9x
Other fields citing it, 2021 to 2024
84 to 205
2.4x
Umami peptides from food
The small protein fragments that give foods their savory taste, and how to find and make them.
Papers, 2021 to 2024
92 to 197
2.1x
Other fields citing it, 2021 to 2024
45 to 59
1.3x
Sixteen picks. Ten broke out.
Run on the map as it stood in 2018, the rule picked single-cell RNA sequencing and MXenes, both of which took off.
The misses teach as much. Generative adversarial networks were the hottest method in machine learning in 2018, and diffusion models overtook them. Neutron star mergers drew a burst of papers after the first one was observed in 2017, and then leveled off.
Growth in papers from 2018 to 2024. A field broke out when it grew faster than nine in ten others.
The map flags momentum, fields growing fast and drawing on more of science each year. Momentum predicts more growth. Whether the growth is worth having is a separate question.
Some of these fields are real shifts in what science can do. Others are booms in a style of paper. Nearly a third of recent oxidative balance score papers name the same US national health survey in their titles. They're on the list because the rule picked them, and we don't edit the rule's picks.
We expect about eleven of the seventeen to break out. Each year we'll post the score, and name the big breakouts we missed.
Telling a real shift from a publishing boom is the judgment our clients pay for.
Every field in the map was scored on how fast its papers grew from 2021 to 2024, and how fast the number of other fields citing it grew. We kept the established fields still rising in 2024, then took the top scorer in each scientific discipline.
We can run the same rule inside your field and show you what's moving fastest and what it rests on.
Ask for a watch list for your field