January 2024 — Monthly Briefing¶
1. What was read¶
906 research articles from 32 of the 33 journals on the roster.
A further 207 records (19% of the 1,113 this month's harvest returned) carry no abstract and are not research: news, correspondence, editor highlights and obituaries. PubMed types them Journal Article and nothing else, so the publication-type filter cannot see them — across the whole index they are 22.7% of it and 35% of general, and 3,801 of them are JAMA. Every count below is over the full harvest and is inflated by this; the figure above is not.
| General | Cancer | Immune | Bioinfo | Sysbio |
|---|---|---|---|---|
| 407 | 189 | 83 | 413 | 21 |
Volume swing against the median month: bioinfo +52%
Check the source before reading anything into this. Two benign causes account for most of these. bioinfo every January carries the Nucleic Acids Research database issue and runs high. And several journals simply deposit nothing in some months: across the 44 months read, 4 of 33 journals are absent from at least one, worst being Nat Mach Intell (20 months missing, 81 papers in total) and Bioinformatics (4 months missing, 2,905 papers in total). This month is missing Genome Res. Queried directly, PubMed returns zero for those journal-months too — the holes are in the source, reproduced exactly by re-running the harvest. So a swing here is not evidence that a field went quiet, and is not usually evidence that the pipeline broke either.
Largest contributors: Nucleic Acids Res 255, Nature 85, JAMA 79, Science 77, Bioinformatics 48, Lancet 41.
2. Net change on the thread curve¶
| Thread | This month | Cumulative | Status |
|---|---|---|---|
agentic-ai |
+1 | +3 / −0 | too-early |
cell-therapy |
+1 / −2 | +86 / −27 | forming |
ctdna-mrd |
+1 / −1 | +25 / −13 | splitting forming → splitting |
foundation-models |
+2 / −3 | +34 / −23 | splitting |
immunotherapy-resistance |
+3 / −3 | +101 / −34 | forming |
interception |
+1 | +61 / −9 | forming |
method-credibility |
−29 | +2 / −158 | crisis |
neoantigen-vaccines |
+2 / −2 | +39 / −17 | forming |
neuro-immune-microbiome-axis |
+3 / −1 | +63 / −8 | forming |
next-gen-checkpoints |
+1 / −1 | +25 / −18 | splitting |
single-cell-3d-genome |
+1 | +26 / −7 | forming |
suppressive-population-inversion |
−2 | +23 / −30 | contested |
Status is derived by rule, not asserted. Fewer than 5 cards → too-early; then refutes ≥ 2×supports → crisis; refutes > supports → contested; supports ≥ 2×refutes → forming; else splitting.
3. The evidence that moved it¶
Written 2026-08-26, in arrears
2024 was added to the corpus nineteen months after the practice started, to test whether its "first appearance" claims survived a lower floor. None of them did. These twelve briefings are therefore written backwards, by someone who already knows what 2025 and 2026 did, and that is the opposite of the position the monthly practice is supposed to simulate. Read them as a baseline for the threads, not as a record of what was visible at the time. 2024 was also read thread-first rather than exhaustively, so its card counts are a floor and are not comparable with 2025 and 2026 month totals.
The agent era starts here, thirteen months before this corpus said it did. 38799228 (Nat Mach Intell) gives a large language model chemistry tools and has it plan and execute syntheses. Until today the corpus dated the first LLM-agent paper to ESCARGOT on 4 February 2025 and built a diffusion-lag argument on it. That argument is void. The tool-using agent is in a tracked journal in January 2024, and 38341654 (GeneGPT) follows in February.
AlphaFold's own limits, stated in its own field's methods journal. 38036854 (Nat Methods): AlphaFold predictions are valuable hypotheses that accelerate but do not replace experimental structure determination. The most successful predictive model in biology publishes its own ceiling in January 2024.
Corrected 2026-08-27
This paragraph continued: "Alongside it, 38082179 reports that
macroscopic resting-state brain dynamics are best described by linear
models — a simple-beats-complex title in the same month. The corpus
previously dated that genre to mid-2025." The paper is real and its result is
as stated: linear autoregressive models fit intracranial EEG and fMRI better
than nonlinear families across 822 participants on predictive power,
complexity and unexplained residual dynamics, and it names four reasons
microscopic nonlinearity is masked at the macroscopic scale. It has nothing
to do with foundation models — no pretraining, no scale, no transfer — and
was put on that thread because "simple beats complex" is a phrase the thread
uses. That is an analogy between two titles, not evidence, and the card no
longer carries foundation-models. It stays on method-credibility, where
the abstract does state a standing assumption and challenge it.
Two neural-immune mechanisms, in the two highest-tier venues. 38134932 (Cell): sensory neurons promote immune homeostasis in the lung. 38085267 adds commensal bacteria driving type I interferon to maintain tolerance. Between them the thread this corpus scores as its healthiest is running on all three of its arms — neural, microbial, lymphoid — in the first month of the window.
Corrected 2026-08-27
This paragraph also cited 37947393, "gliomas hijack adaptive
plasticity to strengthen neuron-tumour synapses", as a third mechanism. It is
not a mechanism and not a January 2024 result: it is a one-sentence Cancer
Discovery Research Watch item on page 13, with no named authors, whose whole
abstract is "Blocking BDNF-driven synaptic plasticity extends survival in
patient-derived glioma xenograft models". It restates 37914930, the
Nature paper this corpus already cards in November 2023 and discusses in
that month's briefing. One study, two cards, and the thread's opening was
credited to the echo. The Research Watch card is now neutral; the primary
paper keeps the finding.
4. Refutations¶
(a) Landing on a tracked thread¶
37874957: recent bendamustine before apheresis degrades CAR-T
outcomes — the product's efficacy is contingent on prior therapy in a way the
label did not anticipate. It is the one thread-relevant refutation this month
adds to cell-therapy that was not already on the ledger.
Corrected 2026-08-27
This section opened on 38071995, "the FDA opens an investigation into
secondary lymphomas after CAR-T", and argued that the platform's safety case
is reopened by a regulator rather than by a trial. The regulatory action is
real; the card is not a second instance of it. The record is a single-page
Lancet Oncology news brief on page 21, one staff writer, no abstract,
reporting the same investigation as 38071978 — the Lancet news brief
one month earlier, already carded and already read as this refutation in the
December 2023 briefing. Counting the echo let one regulatory action move the
cell-therapy ratio twice, and put it in January when it belongs in
December. The January card is now neutral.
38296595 (JITC): no detrimental association between antibiotic use and checkpoint inhibitor therapy, in a cohort assembled to find one. A widely repeated microbiome-adjacent clinical claim fails to replicate, and it fails in the same month the mechanistic microbiome papers above are being published.
(b) Method credibility¶
38150261 (JAMA): a large language model reproduces affiliation bias when reviewing abstracts. The audit of machine judgement begins in the same months as its adoption, not after it — a pattern that holds for every AI modality in this corpus once the window is wide enough to see it.
38172966 benchmarks differential-abundance methods for finding condition-specific cells across multi-sample single-cell datasets.
5. Open items¶
Nothing carried forward.
6. Meetings¶
752 abstracts from 1 meeting.
| Meeting | Abstracts | With full text |
|---|---|---|
| ASCO Annual Meeting 2024 | 752 | 752 |
7. Regulatory¶
| Event | Count |
|---|---|
| trial-first-posted | 569 (of which 521 genuinely oncology) |
| ema-authorisation | 8 |
| fda-approval | 2 |
2 FDA actions — dated from each notice's own prose, not the table's posting date, which disagrees on 7 of 190 rows and once by enough to move an approval into the wrong month:
- 2024-01-12 — FDA approves pembrolizumab with chemoradiotherapy for FIGO 2014 Stage III-IVA cervical cancer
- 2024-01-19 — FDA approves erdafitinib for locally advanced or metastatic urothelial carcinoma
5 EMA oncology authorisations:
- 2024-01-05 — Krazati
- 2024-01-05 — Spexotras
- 2024-01-05 — Azacitidine Kabi
- 2024-01-05 — Naveruclif
- 2024-01-25 — Omjjara
8. Trade press¶
46 on-topic of 286 items across 3 feeds (n-stat 272, n-aacr-blog 10, n-acir 4). Headlines are selected by subject keyword, not recency — most of what these feeds carry is health policy and industry finance the corpus does not read.
- 2024-01-31 — Experts Forecast 2024, Part 4: Cutting-edge Tech for Oncology Drug Discovery —
n-aacr-blog - 2024-01-31 — We’re naming cancers all wrong, oncology leader says —
n-stat - 2024-01-30 — Experts Forecast 2024, Part 3: Precision Medicine for Cancer Treatment —
n-aacr-blog - 2024-01-30 — STAT+: Ultima Genomics launching high-end DNA sequencers that can read genome for $100 —
n-stat - 2024-01-26 — AACR Journal Spearheads Trendsetting Blood Cancer Conference —
n-aacr-blog - 2024-01-25 — STAT+: Early detection of pancreatic cancer could get a boost from AI that identifies people at higher risk —
n-stat - 2024-01-25 — HPV vaccine study finds zero cases of cervical cancer among women vaccinated before age 14 —
n-stat - 2024-01-25 — Two genomics powerhouses team up to expand infectious disease surveillance —
n-stat - 2024-01-25 — STAT+: What to make of Gilead’s Trodelvy troubles and Elevation Oncology’s surprising rise —
n-stat - 2024-01-25 — ‘The response defies belief’: Year of inaction leaves children at risk from bad cancer drugs —
n-stat - 2024-01-24 — STAT+: Pharmalittle: We’re reading about Lilly’s gene therapy for hearing, Shkreli’s life-long ban, and more —
n-stat - 2024-01-24 — STAT+: Lilly’s gene therapy restores hearing in an 11-year-old, paving way for more treatment —
n-stat - … and 34 more on-topic items.