brain — overview · 2026-10-02

When	Item	Actions
No tasks due today

Time	Slot	What
10:00–12:00	Other	SW/AI Development meeting (Zoom)
13:30–14:00	Admin	Cleanup (optional, pre-strategy)
14:00–17:00	Strategy	Strategic (with Claude)

Day	Item	Actions
No deadlines this week

Who	What	Sent	Status
JJ	Step B — beat touchups on the ~50-case validation cohort.	Step A sent 2026-05-19	parked
Uwe Nixdorff (EPC)	Mutual, agreed 2026-08-30: EN supplies a template for Uwe's LÄK ethics application, Uwe supplies 20 cases up front.	agreed 2026-08-30	to-send
Thomas Koeck + Katrin Göpfert (Mainz)	Auftrag-type contract, OLINK proteomics — 880 samples, 3 Target-96 panels, €110,035.13.	2026-06-01	awaiting
Sadiyya Farooque · +Carrillo (Einkauf, Abt. 3.1), Lotz (external, LV author)	Run the Angebotseinholung for the KI-Körperzusammensetzung tool (~€98.4k net) and obtain the candidate-supplier list from Lotz.	asked 2026-09-04	running
Raphael	Retraining + data/model versioning + Metrics-Reloaded pipeline; nnU-Net Track-1 SAX baseline.	2026-06-05	running
Simon (Bohlender)	Data send/receive into RADICAL while scanning, latency-aware (WP3 I/O).	discussed 2026-06-28	running
Raphael	UKB Cat 523 re-download.	2026-05-12	parked
Sadiyya Farooque	Forward the Birkholz answer on >€50k procurement + ask EN for the case basis.	2026-06-02	parked
Deniz	Get a second EUDAMED LAA/LUA user set up so the Radical device (Basic UDI-DI 426256336RadicalJ7, GTIN 4262563360007) can be registered — EN is sole user without register rights.	2026-09-30 (e-mail sent)	awaiting

Hannah

On her plate (owner): KW38 (she reported the short week 15.–17.09.) — Konfliktliste von Quang: Impression-Datenkonflikte lösen (7 h, in Arbeit); Akten suchen und sortieren nach einer Unstimmigkeit in den Patientendaten (7 h, Raphael hat ausgeholfen); Impression-Terminierungen (dauerhaft in Arbeit, wartet auf Rückmeldungen); Aufarbeitung nach Urlaub ✓ (7 h, Frist 15.09. gehalten); Impression-Mappen pausiert — der Großteil liegt bei VP zum Prüfen der CMR-Reports.
Waiting on you ← (indirect, but it is the OLINK chain): Blut für Celltrend und OLINK heraussuchen und vorbereiten is pausiert, geblockt durch VP since 2026-08-20 — 29 days, with the comment "Rückmeldung ausstehend bzgl. T1/Impression". Her step is the physical sample handling; yours and VP's is the sample definition, two days overdue. Same chain, blocked at both ends. Separately: her Weiterbeschäftigung sits on your desk per SF's KW36 report.
Cadence: KW38 is her first return since KW34 — she was on leave in between, which the KW38 sheet accounts for.
Nudge: light.
(note: Hannah = HH; not QAH. QAH = Quang An Hong — the Claudia how-to is Quang's.)

works best with

routine, steady cadence; occasional reminder is enough. Low-maintenance.

Tilman (TR)

On his plate (owner): KW38 (~22 h, template followed cleanly) — Myoflame-19 Zentrumsvergütung (Kontrolle Abrechnung, Freigabe, Prüfung — in Arbeit, 2 h, fixe Deadline 2026-09-21, Monday); Impression Amendment V6 (lesen, korrigieren, Info-Pamphlet schreiben — in Arbeit, 7 h, fixe Deadline 2026-09-25); T1/Impression-Ordner-Korrektur (in Arbeit, 5 h — the same folder mess Hannah is working from the other side); Institut-Bestellungen ✓ (SAP APM, Formulare, 3 h); HiWi Reinstein onboarden ✓ (5 h — the onboarding is done even though the Einstellung is still blocked on you).
Waiting on you ←: — (nothing blocked). Myoflame board item c2 is CLOSED as of 2026-09-02 — EN decided the 13 grants-office questions himself rather than wait; TR never answered it. Do not re-raise c2. His last delivery was the Myoflame-19 Apotheken-Ist-Zahlen (27 Aug, absorbed same day) plus a KW35 status (28 Aug). Nothing of yours sits with him right now.
Nudge: reminders help a lot — surface his items the morning before you see him.

works best with

clear asks + reminders — he's just-in-time and gets done what you tell him; a nudge lands well. Ask him for a short structured weekly status back (also feeds the month-of-Cowork efficiency push you gave him).

Deniz

On his plate (owner): EUDAMED second LAA/LUA user for Radical device registration (asked 2026-09-30, 20-company/product/radical).
Nudge: light.

works best with

hands-off; delivers reasonably well. Visibility into actual load is low, but output is fine.

Simon

On his plate (owner): KW36 (~33 h; nothing filed for KW37 or KW38) — Refactor DICOM storage: new image-database models, implementation started (in Arbeit, 5 h); QM-Implementierung ✓ — the QM report is now responsive and displayable (10 h); EOS monitoring ✓ (6 h) — multiple problems this week, the hard drive was full; improved monitoring stack (in Arbeit, 12 h) — the stack had been unstable "for some time now". Plus the standing RADICAL-latency feasibility for imaging-4 WP3 I/O.
⚑ Worth one glance: two of his four KW36 items are EOS/monitoring firefighting, not project work, and the disk-full event is the kind of thing that costs a week silently. He also blocks Luisa on a code push (since 2026-09-11) — she is waiting on his segmentation-quality rules.
Nudge: low.

works best with

give him a problem, get out of the way. Self-assigns and stays well aligned.

Quang

On his plate (owner): KW38 — and it is a week of things closing. Myoflame CPET → in progress, now blocked only on permissions; Impression CPET → done (vorläufig) (was "in progress" in KW35); Model-Selection-Strategy für schnelle Iteration → done (vorläufig); Mapping Pixel Values → LGE-SAX annotation support added (pixel-value extraction still to come); GoetheCVI → bugfixes + QOL, tag sorting and filtering in the library; new: Impression Birth-Date-Konflikte auf EOS (~5 cases left, assisted by Hannah and Til) and REDCap Date Conflicts (same pair helping). Carried: Data Management, InBody, LGE project, LA-Experiment für Alexandros, VHR.
Perfusion — handed to Neel on purpose, not paused. His KW38 says it from his side: "Neel: perfusion — currently developing deconvolution algorithm" and "Perfusion — supporting neel from the sidelines." The KW35 "on hold / wegen der Research Sachen" was the step before the handover, not a third abandonment. EN 2026-09-18: "QAH is doing good things — as always." The coverage-risk line below stays as a watch, not as a finding.
The only soft spot is labelling, not reporting. He files, but reuses the same template file without re-dating the title cell — which is why KW38 first read as a KW35 duplicate. One edit of cell A1 each week removes the whole problem.
Coverage risk (important-but-neglected — track these, not his flow): perfusion follow-through; closing loops with SOPs (e.g. LAX quantification); his MD-thesis — would open strong grant-writing.
Guides: Neel.

works best with

hand him the important problem; he'll often overrun the assignment on things he cares about (made the software ~5–10× faster). Lighter on what doesn't engage him.

Raphael

On his plate (owner): KW36 — a single line, and it is a good one: Training Pipeline Preliminary Production Setup — Fertig (polishing + Model-Selection interface for GoetheCVI, 32 h, "For now done"). The self-block that ran from KW27 to KW34 ("Geblockt durch: Raphael") is cleared in this return. Expect KW38 in his first week back; there will be no KW37.
Back since Mon 21 Sep — day 4 after leave 04–18 Sep 2026. 03 Sep was his last working day and EN used the window (ticked on the focus page). His KW36 return is the most recent on file (KW29–KW36 all present, 50-team/raphael/weekly-status/); expect KW38 in his first week back, not a backfill. Two delegations go live with him again — retraining / versioning / Metrics-Reloaded (since 05 Jun) and the UKB Cat 523 re-download (still blocked, UKB offline, do not nudge). Contours are company priority #1 and the sell-blocker, so the first conversation is what he picks up first.
Guides: Luisa (under review — her KW29 next step is gated on his review; fresh evidence for the fit question).

works best with

bring the analytical scaffolding to discussions — picks things up and delivers, but is non-analytical in problem-solving; currently the weakest link in the engineering group. Open question: is he the right person to be guiding Luisa?

Luisa

On her plate (owner): KW38 (~22 h) — MLflow/GoetheCVI: Erweiterung für neue Datenquellen und 3D/4D-Daten (Trainings- und Tracking-Pipeline auf mehrere ZIP-Quellen und Base-NIfTI erweitert; 19 h, in Arbeit — End-to-End-Test steht noch aus, reale Mapping-T1/T2-ZIPs noch nicht durch); Kontur-Lücken in Trainingsmasken schließen (3 h); Medizinische Evaluation / Qualitätskriterien Segmentierung — GEBLOCKT durch SB (Code-Update) seit 2026-09-11 (sie hat die Painchaud- und Mariscal-Harana-Recherche an SB weitergegeben und wartet auf seinen Push); Evaluation Sample-Limit (#10) pausiert zugunsten der priorisierten GoetheCVI-Aufgaben.
Waiting on you ←: — . Her gate has moved from RJ to SB (since 2026-09-11), which is a real change: the guidance-fit question below was framed around Raphael, and the actual dependency now runs through Simon.
KW29–KW38 unbroken — the only complete series in the group. Hours filled every week, blockers named with dates. She is the model return.

works best with

she's very good and early in her career; thrives where she can grow. Currently guided by Raphael — guidance fit under review.

Neel

On his plate (owner): KW38 — one line, 20 h: Perfusion experiment — figuring out algorithm for deconvolution (in Arbeit). Nothing else listed.
Guided by: Quang — and the perfusion assignment is deliberate. Quang's KW38 confirms it from his side: "Neel: perfusion — currently developing deconvolution algorithm", "Perfusion — supporting neel from the sidelines." EN 2026-09-18: Neel is on perfusion on purpose. (An earlier read of this card flagged possible drift; it rested on Quang's stale KW35 sheet, before his KW38 was correctly identified. Not a finding — do not re-derive it.)
Flag: he left the week field as the template placeholder <KW / 2026>; the file was filed as KW38 from its filename (..._14_sept). His MDR software-test automation, which was the KW29 plate, has not appeared in a return since.

works best with

capable; takes on larger projects (e.g. automating the software tests required for MDR). Output is filtered through Quang, so completion visibility is partial.

Anna-Katharina (AK) — **leaving soon** (EN 2026-09-21, no date) · profile TBD (first return KW29)

Leaving "soonish" — handover open. Four clinical tasks (CPET, EKG, BIA, data export) pass to AM, who is currently only the deputy; the Lehrveranstaltung has no named deputy at all; "Review current report" is flagged hanging; the Rotterdam abstract has no stated owner after her. Node: 50-team/anna-katharina/STATE.md.
Werkstudentin ("Anna" in SF reports); roster: not in the "Geblockt durch" dropdown (can't block anything). Actual work was flagged "unclear" — KW29 return reveals it: paper- and teaching-heavy, run largely by VP. Working-style profile pending Eike.
On her plate (KW34 — her last return; silent for four weeks): SOP CPET Korrektur (Korrekturen von Valentina einfügen, in Arbeit); SOP CPET Layout (Vorlage von Til erhalten, in Arbeit); SOP Blutabnahme/Labor Layout — verschoben, "Hannah sucht das Word-Dokument nächste Woche raus"; Impression Probandenmessungen (Ruhe-EKG, InBody, Spirometrie, Ergospirometrie, Blutabnahme, 8 h); Probandendaten exportieren (CPET, InBody, MRT) — with the open question "Wo soll ich die Daten eigentlich speichern?", unanswered; Myoflame-2.0 block with Til (Kontakte, DocuSign, NDA rausschicken, Dropbox, Texte, ~9 h); Montags-Team-Meeting.
⚠ Two things. "Wo soll ich die Daten eigentlich speichern?" is a study-data-storage question from a Werkstudentin that has sat unanswered since KW34 — that is a governance question, not an admin one. And the ESC-Munich items that dominated her August plate are moot: ESC has passed, and the question you settled on 15 Sep was whether that prep was HiWi (paid) or Doktorarbeit (unpaid).
Waiting on you ←: — (the SF/ESC-travel block from June is spent — ESC is over).
Flags: hours missing on almost all items; mis-used "Geblockt durch: mich" (self) on the two papers (not a valid blocker); scope reads above typical Werkstudentin level. One impossible date ("3036-08-03").
Nudge: TBD — confirm role/goal with VP + Eike.

Alex

On his plate (owner): DECIPHER LA + LAX analysis — returned/done 2026-07-16 (briefed at kickoff, dataset access handed over).
Waiting on you ←: — (kickoff done).
Nudge: onboarding — set expectations at the first meeting.

works best with

TBD — scaffold created 2026-07-09, fill at first meeting.

Valentina

On her plate (owner), the first two added 2026-09-13 from EN's 12 Sep forwards, the third added 2026-09-18:
DGK Zusatzqualifikation CMR — Rezertifizierung als Stätte. She wrote to DGK on 03.09. (Priorität hoch) that "Prof. Nagel und ich besprechen den Plan und werden Ihnen dies schnellstmöglich mitteilen" — that reply does not appear to have gone out — fifteen days on (03 Sep → 18 Sep). The certificate expires 19.11.2026 (62 days) and the only blocking step is a signed Antrag. She also asked DGK to delete the dead @kgu.de address. → 05-admin/2026-09-12_DGK-Rezertifizierung-Staette-CMR.md. EN marked it "to discuss with VP", so this is a joint item, not a handoff.
~~Silvio Baccaro / EACVI Research Grant~~ — returned 29 Sep: Silvio declined (fellowship elsewhere). Her review/forms step falls away.
OLINK-Samples definieren — joint EN + VP, due 2026-09-16, now two days overdue. Slot-Budget, Legacy-Aliquots, 880 vs 968, Bridge vor Randomisierung. Source 10-research/olink/STATE.md:97, authored by EN on 15 Sep for "morgen". This is the live OLINK item; the Mainz contract is not — that half sits with Koeck/Göpfert in signature and is parked to 01 Oct. The distinction matters because the contract row is the one that keeps drawing the eye. Added here on 18 Sep: it had been carried on the project axis since 16 Sep but never reached a person card, which is how a joint item goes quiet.
(the Alcedis reconciliation closed 2026-08-18: the 12:00 meeting answered all 19 questions and settled Pos 3 as site money. Do not re-derive it as an open send.) Otherwise see 50-team/valentina/STATE.md.
NEW, from the weekly returns read on 2026-09-18 — three people are blocked on VP right now. These are not delegations EN made; they surfaced from the team's own sheets: Hannah's Blut für Celltrend und OLINK, blocked on VP since 2026-08-20 — 29 days (comment: "Rückmeldung ausstehend bzgl. T1/Impression"); Sadiyya's DTA Puntmann/Müller, with VP for review; Sadiyya's EDV ERC Faculty Form (Valentina), at the Dekanat. Hannah's is the one that matters: it is the physical end of the same OLINK chain whose definition is two days overdue on the joint EN+VP item above. The chain is blocked at both ends, by the same two people. Also on her desk from Hannah's sheet: the Impression-Mappen are pausiert because the bulk sits with VP for CMR-report checking.
Waiting on you ←: — (what came back from that meeting was yours, not hers: D2 settled 08-21, C1 answered 2026-08-24 — the €500k IMP line survived the adapted costing unchanged. Nothing left on either side of this thread.)
Nudge: align, don't task-track.

works best with

discuss and agree content first — and emotionally — before she runs. A power-house who brings things into action fast; very quick to see that something doesn't work, slower to find the fix; dislikes ambiguity / grey values. Without agreed direction she can run full-speed the wrong way. Give her a clear ruleset to operate inside rather than open-ended todos.

Sadiyya

Status: missed 5–6 Monday management meetings; asked for a weekly activity list → off sick ~2 weeks → back 2026-06-22, declined the management meeting citing inbox load. Prior read: likely past repair; replacement in motion (HiWi + TR).
KW26 update (2026-06-29): delivered the requested weekly activity list — 35 h, substantial: Finanz-/Personalkostenplanung 2026 complete, >100-mail inbox cleared + digital filing restructured, CMR End Letter, CMR-Retreat coordination, Deniz Reisekosten prep, FK-Z for VP's SE, VP Projektvertrag → ready for Leistner signature then Drittmittel. Caveat: much is "in Vorbereitung / zur Unterschrift bereit" (teed up, not closed) = the E9-coordination tier the HiWi+AI replacement targets. One strong week (and the first visible one) ≠ overturned pattern → watch 3–4 weeks before re-rating.
HiWi Mehrkosten / Umbuchung (Imaging 4.0): decided 2026-06-29 (unblocked — was her blocker). VP contract + Deniz Reisekosten just need downstream signatures.
Contract ends 12/2026 (confirmed 2026-06-23) — ~6 months runway. Structural call stands (role is HiWi+AI-replaceable); the strong week shifts the human handling toward using the runway well, not an early adversarial exit.
DECISION — how to proceed (use-up-until-12/2026 vs negotiated early exit; align HiWi + TR replacement to the 12/2026 cliff): 30-min session — now schedulable, now better-informed by KW26.
On her plate (owner): KW36 — 15 items, and 10 of the 15 are marked "Geblockt", which is the single most informative line in this whole refresh.
Blocked on EN (2, was 3): ~~HiWi Raphael Reinstein Einstellung~~ — CLOSED 2026-09-21, he is employed; Weiterbeschäftigungen allgemein ("in Klärung mit Prof. Nagel"); Weiterbeschäftigung Hannah Horn (running via Studienzentrale).
Blocked on VP (2): DTA Puntmann / Müller (bei VP zur Durchsicht, parallel an Legal); EDV ERC Faculty Form (Valentina) (beim Dekanat FB16 eingereicht).
Blocked on external partners (5): ME/CFS Drittmittelantrag (Abteilung genehmigt, Fördervertrag muss noch); MYOFLAME-19 Substudy Fördervertrag (Rechtsabteilung geprüft, Entwurf liegt); JGU Mainz Forschungsvertrag (geprüft und nach Mainz geschickt — this is the OLINK contract half); CellTrend Dienstleistungsvertrag (von Dr. Heidecke inhaltlich freigegeben); DTA Deutsches Herzzentrum/Charité Berlin (liegt in Berlin).
Moving: Beschaffungsantrag Focus Group HFpEF (erstellt, an den Einkauf — the Carrillo procurement); Dienstreise Hamburg/DZHK (DB-Rechnung da, Hotelrechnung fehlt); Freigabe Auslandsdienstreisen durch FMM ✓; Prüfung Verwendungsnachweis Imaging 4.0 — "Raphael Jöckel E12, doppelt besetzte Stelle".
⚑ That last line is the FLOTO answer. Her KW36 entry names the E12 question directly — it is the same [GAP] the Sulzmann draft cannot close. Worth reading her line before writing yours.
Ten of fifteen blocked is the pattern to read, not the count. Three blocks are yours and all three are personnel — one conversation clears them.
Waiting on you ← (2026-09-27) Urlaubsantrag 12.–16.10. und 11.–20.11.2026 — sent 25 Sep, fwd by EN 26 Sep without disposition; filed in 50-team/sadiyya/STATE.md ## Inbox. EN approves or declines.
Waiting on you ←: Raphael Reinstein contract "zur Freigabe bei Prof. Nagel" — framed as waiting-on-EN (same pattern as KW28; EN typically turns these around next day). NEW, untracked by SF: AK is blocked on SF for ESC-Munich travel (train ticket, hotel-cost check, ticket-cost submission) since 2026-06-28 (~3 wks). 2026-07-23: SF sent per-Punkt status of the DZHK Imaging 3.0 VN Rückfragen (2–5, Frachtkosten, Reisekosten), flagging where EN is needed → EN review + input before the 28.07 gate. 2026-08-03: EN submitted the VN/Schlussbericht 81Z0200109 to DZHK/FMM — CLOSED on our side; do NOT resurface as awaiting.
DZHK Rückfragen to Verwendungsnachweis/Schlussbericht 81Z0200109 (Imaging), delegated 2026-07-18 — submitted to DZHK/FMM 2026-08-03; CLOSED.
OLINK follow-up — OFF her card as of 2026-09-12. EN took the Mainz chase over himself ("the relay through SF cost the same as the call"), so SF is no longer in that loop and the question of whether her chase ever went out is void, not overdue — do not re-derive it as an open ask on her. The node still reads "in signature" and the row now sits under EN's own name in DELEGATIONS.md row 3. (This card carried it as an open loop from 03 to 13 Sep; corrected in the 13→14 Sep regen.)
Vertragslaufzeiten — delivered 03 Sep. EN asked for a list of all staff contract end dates; she sent the Personalplanungsübersicht (Jahresplanung_2026_final.xlsx) the same day and confirmed "Die Zeiten stimmen". Nothing outstanding on her side; the remaining step is EN's own conversation with VP. The workbook stays in the mailbox, not the brain (staff personnel data).
KW36 delivered (2026-09-04) and now answered back. Weekly status xlsx + Wochenbericht PDF, both now in 50-team/sadiyya/weekly-status/. EN transferred her report into the template himself and asked three questions; she answered the same afternoon (16:12) and accepted the template for future weeks — filed as KW36_SF_Rueckfragen-EN_2026-09-04.md. Substance: the ME/CFS items 1 and 2 are two different releases (internal, by Stahlhut; and the draft contract, by Vural, to the external partner); the Studienzentrale ended 12/20 and lives on only as a Durchgangskonto, so any report line treating it as an active project is noise. She is the most consistent returner in the cadence. Second week running, though, the report needed EN's own hand to become readable — whether the next one arrives usable is the thing to watch, not something to chase.
⚠ KW38 read line-by-line against KW36 (EN, 2026-09-18: "the 'in Arbeit' is a full week's work…!!"). He is right, and the diff says why.
Four of the six "In Arbeit" rows carry comments that are verbatim identical to KW36, two weeks earlier. Beschaffungsantrag Focus Group HFpEF — "Antrag erstellt und an den Einkauf weitergeleitet", word for word. DTA Charité Berlin — "Interne Prüfung abgeschlossen; Dokument liegt in Berlin", word for word. FLOTO/Verwendungsnachweis — "Infos herausgesucht: Raphael Jöckel E12…", word for word. Hamburg-Buchungen — "DB-Rechnung liegt vor, Hotelrechnung noch ausstehend", plus four words. Nothing moved on any of them; the label moved.
Two of those rows were "Geblockt" in KW36 and are "In Arbeit" in KW38 while still naming a blocker — DTA Charité (Geblockt durch: externer Partner) and DTA Puntmann/Müller (Geblockt durch: VP). An item cannot be her work-in-progress and someone else's ball at the same time. Re-labelling stasis as activity is the specific thing to name.
Zero "Fertig" in KW38 (KW36 had one). Zero hours filled in either week — ~ Stunden diese Woche and ~ Stunden insgesamt are empty on all 15 rows, and Geblockt seit is empty on all of them too, although the template header says in so many words: "Geblockt = Name, Inhalt, Datum." The two columns that would let anyone check a week's work against a week are the two she never fills.
Hamburg (DZHK) carries a fixed deadline of 9.9.2026 — nine days past, still "In Arbeit", still waiting on the same hotel invoice.
HiWi Reinstein is still "Geblockt" with the blocker field now blank, while the comment says onboarding is running on site — and TR reported the onboarding as ✓ done in his own KW38. The row is stale on her sheet.
What genuinely moved, and it should be said: CellTrend advanced properly — legal clearance from Frau Vural obtained, DocuSign signature process started, signatories named (Tilman, DL, VP, Heidecke, Vural). The KW36 (Name eintragen) placeholders are now filled with real people (Göpfert in Mainz, Thiede in Berlin). Two new items appeared (Claude scientific account; SAP-APM Antrag). That is perhaps a day, framed as five.
Three items name EN as blocker, two marked ASAP — but they are not the same kind of thing. Two are stamps: Weiterbeschäftigung Hannah Horn (Antrag + Sachgrundschreiben erstellt, liegen bei EN) and the SAP-APM KST-Freigabe. EN signs, she forwards, done. One is a real decision: Weiterbeschäftigungen (allgemein) — EN 2026-09-18: "if I need time, thoughts, strategy, meeting with staff, etc to work out the optimal and payable workforce for next year…". Marking that ASAP is a category error on her sheet. It is not chaseable, it is not administrable, and no amount of follow-up from her produces it.
Correction, EN 2026-09-18: an earlier version of this card read her thin week as partly caused by lines terminating at his desk. He rejects that and the evidence does not support it. He turns the stamps around; the external blocks (ME/CFS Foundation, JGU Mainz, Berlin, Vertragspartner) are real and cannot be defined away; and the one genuine EN-block is a strategy item, not a delay. Counting blockers is not measuring turnaround, and this card made that mistake once.
★ The Claude-account row is the whole pattern in one line (EN, 2026-09-18). Sequence, from the mail chain: EN asks 15 Sep 16:50 ("können Sie bitte herausfinden, ob die Universität einen Science Account für Claude Anthropic hat. Hochschulrechenzentrum sollte das wissen."). She forwards two words of her own to HRZ on 17 Sep 15:53 — a two-day lag on a one-line forward. HRZ answers 18 Sep 09:36. She forwards the answer on. Total work: two forwards. And on the KW38 sheet — which arrived at 10:06, half an hour after the answer landed — the row reads "Claude scientific account · In Arbeit · Cluade Scientific Account angefragt": a status that was already obsolete when it was typed, with the product name misspelt. EN: writing the row took longer than doing the task. The answer itself is filed at 40-infrastructure/claude-dsgvo/STATE.md ## Inbox.
What that sharpens. The problem is not load and not blockers — it is that the sheet is being filled as a record of having been involved rather than as a record of work done, so effort spent on reporting scales with the number of rows rather than with the work. That is also why the hours columns stay empty: hours would immediately expose the ratio. The escalation mechanism recorded at the end of this file fires on silence; this is the other failure mode, visible activity with no delivery, and it is not caught by that mechanism.
KW38 filed 2026-09-18, on EN's call. Her return arrived with the title cell still reading "Sadiyya Farooque · KW 36 / 2026" — a sheet carried forward without re-dating. EN: "as always — too little thought — just file the new one as KW38." Filed as KW38_SF_Weekly-Status.xlsx; _index.md records the discrepancy. Standing rule from this: for SF a stale title cell is a copy-forward, not a re-submission — file by arrival week, do not ask.
"Doppelt besetzte Stelle" (Imaging 4.0 VN) — EN 2026-09-18: Linda will sort it out. Her line "Raphael Jöckel E12, doppelt besetzte Stelle" is therefore not an open item on her card and not something EN needs to chase. It does not touch the FLOTO [GAP] on whether an E12 was ever applied for — that one is still EN's alone.
Nudge: Umbuchung decided 2026-06-29 → schedule the 30-min proceed-session.

Raphael Reinstein (RR)

Employed — confirmed by EN 2026-09-21. The Einstellung had been sitting on EN's desk since an Einstellungszusage dated 15.09.2026; it is done. Two rows on SF's sheets that still read blocked on EN are superseded.
Onboarded by TR, who reported it ✓ done in his KW38.
On his plate (owner): not yet recorded. One column of 50-team/Hiwi-job-list.docx is his; the X / XX / xx legend there is defined nowhere.
Working alongside SF on administration and finance as continuity cover (EN, 2026-09-21). Context in his node.
Waiting on you ←: —
Node: 50-team/raphael-reinstein/STATE.md

works best with

not yet observed — fill at the first touchpoint.

Aynaz Mousavi (AM)

Added to the roster by EN 2026-09-21. Until today she existed in the brain only as a column heading in Hiwi-job-list.docx, which is why that file could not be routed for nine consecutive mornings.
On her plate (owner): not yet recorded; one column of the HiWi grid.
Waiting on you ←: —
Open: tier, who guides her, contract dates, whether she joins the weekly cadence.
Node: 50-team/aynaz/STATE.md

works best with

not yet observed — fill at the first touchpoint.

Note about people (autosaves; send when ready)

Proposed for next Fri 14:00: (set Friday topic in STRATEGY.md)

Work next — one per lane

Next grant

NEXTDFG2 — RT-uncertaintyhigh

**Lane occupant, set by EN 2026-08-29.** Node: `10-research/grants/rt-feedback-cmr/` — *RT-feedback CMR, trustworthy automated cardiac MRI with human oversight*. EN calls it **"dfg2 — rt-uncertainty"**; the working folder is `dfg2_20260829/`, so that is the name in use. Currently **drafting v3** (2026-08-29): panel widened beyond LVEF to LV mass, RV and the atria, unified by the base/AV plane; uncertainty attributed anatomically; reduction as an action space.
**This displaces GREY-ZONE**, which the grants node still marked "★ priority grant, next to write" from 2026-08-09. That marking is superseded — see the grant-lane rank 2 below.
**MYOFLAME-2.0 is ahead in time but is NOT this lane.** EN 2026-08-29: *"Myoflame is theoretically ahead, but it's now in execution stage."* Remaining on it: **recalculate the budget and read it once more — Monday**, then submit by the 02.09. Skizze deadline. Execution, not a writing choice.
**Feeding it:** the HFpEF reproducibility data discussed 2026-08-29 is for this grant and EN flags it **urgent**.

GREY-ZONE (deferral on converged uncertainty)medhard deadline

**Demoted from rank 1 by EN 2026-08-29** in favour of DFG2. Was marked "★ priority, next to write" on 2026-08-09.
The node itself already carried the doubt: *"Active drafting has moved to the RT-feedback node — revisit whether this deferral grant is still a parallel submission or on hold."* That revisit is now the open question for this lane slot. Marinov/KIT engine, DECIPHER PV-loop reference.
The funding ladder — €2.5M survival (2031–40) / €10–20M ambition. EN = AI-imaging, VP = trials.
Assembly started 2026-06-05 — keep momentum; longlist sequencing + DFG HFpEF Methods rewrite. Submission dates are hard.
**(2026-06-11)** New collaboration signal to fold in: **David Bode (DHZC)** — 120-pt transmyocardial + coronary-sinus + femoral paired sampling, invasive hemo + blood gas, HFpEF-rich, banked aliquots; collaboration-only (not a biobank). Strong wet-lab counterpart to leg-stress CMR energetics → hook to **DFG cmr-dyspnea-energetics** + **OLINK**. Otherwise next when time: **work the DFG grant**.

Next paper

— open —

No rank 1 in this lane. The item below is in it, but is not marked as the next one — and it is not promoted for being alone here.

Papers — nothing writable yetmed

**The lane is EMPTY by EN's own call, 2026-08-29: *"next paper is currently empty. I have no data lying around."*** Deliberately not filled. This card holds the candidates so they stay visible without being promoted.
**Closest to a paper — the voxel map** (CONSEPT/MPE). EN: *"probably the closest to a paper, it just needs a bit more structure."* It has its data: n=100 embedded, clean n=94, septal T1 1141±44 ms / T2 38.9±2.9 ms, within-case CV median T1 3.4%, both tools built (`mpe_voxel_explorer_n100.html`, `mpe_binned_lattice_n100.html`). *Naming: EN says "voxel map", the node says "mapping pixels / MPE" — worth settling on one word before it becomes a title.*
**HFpEF reproducibility data — urgent, but it is data work, not the paper.** Discussed 2026-08-29 for DFG2; EN flags it **urgent**. It feeds the grant first.
**P3 LGE multireader has LEFT this lane.** EN 2026-08-29: *"LGE requires data so it actually (no matter what i said) in the data acquisition queue, together with atria in VHR and in HFpEF."* Superseding his own 2026-08-13 "front of the papers queue". Now tracked in `10-research/data-acquisition-queue/`.
**(2026-08-10, Eike — pipeline gap):** OLINK deprioritised (too slow/costly, not worth it), subclinical/provocation paper closed, and PE / DEC-I / DEC-C all on long fuses → **no NEXT paper**. Strategy-session task: think through what is genuinely publishable from the data we ALREADY hold, without new patients.
**(2026-08-13, three days later) a concrete answer to this already exists and was not folded back here:** the 4-paper plan at `10-research/consept/papers-plan_3-paper-todo_2026-08-13.md` — P1 DECIPHER reproducibility (safe win, doubles as the RT-feedback DFG Vorarbeit), P2 group discrimination (riskier), P3 LGE multireader Delphi (parallel, EN + Sven Plein), P4 automated LGE vs the Delphi consensus. All four run on data already held. Whether that closes this initiative is EN's call at a Friday block.

Next collaboration

NEXTNixdorff + Kelle — both met Sun 2026-08-30highOVERDUE (Sun 30 Aug)

**Lane occupant, set by EN 2026-08-29: *"collaboration: Uwe Nixdorff and Sebastian Kelle (I am meeting both tomorrow)."*** Both were already on his 2026-07-03 partner list — **Uwe Nixdorff (EPC)**, whole-body CMR, ~5000 + health evaluation + follow-up; **Sebastian Kelle (Berlin)**, CMR. *(The partner list spells it "Nixdorf" with one f; EN writes "Nixdorff" — his spelling used here.)*
**Neither meeting is on the brain calendar.** Sunday currently holds only the Sven Plein slot at 15:00 — and that one twice.
Per the Cooperations scope, go in with the four things named: what the goal is (data / grant co-app / IP / methods / co-authorship), what EN gives, what he gets, and the push-timing.

Cooperations — the frameworkhigh

The framework behind the lane: goals + how to organise partnerships, per partner. **SCOPE (EN 2026-07-03):** partnerships for **his own research**. Inbound *via* the HFpEF focus group or ageing (e.g. David Bode / DHZC) is research-politics, NOT a cooperation.
**Full partner list (2026-07-03):** Berlin — Sebastian Kelle (CMR) · Berlin — Anja Hennemuth (AI) · EPC — Uwe Nixdorff (whole-body CMR) · HZ Bad Bevensen — ??? (CMR) · Messina — Tommaso + Claudia (CMR: HR-cine) · Heidelberg — Florian Andre + Benjamin Meder (cardiomyopathies) · + more TBD. **Confirm keep/drop:** Bender/Loewe IBT (AI). Scaffolding at `10-research/admin/research-politics/cooperations/`.
**SCOPE (Eike 2026-07-03):** partnerships Eike wants for **his own research**. Inbound *via* the HFpEF focus group or ageing (e.g. **David Bode / DHZC**) is research-politics, NOT a cooperation — do not raise it here unless Eike flags it. (Bode moved to the focus-area track.)
**Eike's partner list (2026-07-03):** Berlin — Sebastian Kelle (CMR) · Berlin — Anja Hennemuth (AI) · EPC — Uwe Nixdorf (whole-body CMR, ~5000 + health eval + follow-up) · HZ Bad Bevensen — ??? (CMR) · Messina — Tommaso + Claudia (CMR: HR-cine, other) · Heidelberg — Florian Andre + Benjamin Meder (CMR: cardiomyopathies) · + more TBD.
**Prior-listed, confirm keep/drop:** Bender/Loewe IBT (AI) — was on the mid-June list, not on Eike's 03-Jul list. Scaffolding at 10-research/admin/research-politics/cooperations/.

⛔ blocked by: per partner — goal (data / grant co-app / IP / methods / co-authorship), what Eike gives, what he gets, status, push-timing.

Next private-praxis endeavour

NEXTBad Orburgentself-imposed

German consolidation step — pulls Beyond + medneo + Am Turm into one place; local doctor runs the FFM back-office.
Generates revenue + de-risks the uni exit; bridges to the Mojácar longevity centre. Next: Eike's concrete move after the Strauss/Adel chats.
**Sequencing (Eike 2026-06-07):** do Bad Orb FIRST — asap with VP (next few days).
**(2026-06-11) Eike + VP executing directly this weekend (Sat/Sun)** — handled outside the Friday strategy block.

Next company priority

NEXTContours (CMR-AI segmentation)high

**Lane occupant, set by EN 2026-08-29: *"keep 'contours' as the #1 until moved."*** Moved out of `actioned` — it had been parked pending the Luisa CorSeg verdict and Raphael's nnU-Net baseline; EN has now made it the standing company priority regardless of those triggers, until he moves it himself.
**Model verdict, EN 2026-08-29: *"the verdict from all the models is: ours is better so we keep training the unet."*** So the two-track call stands and the nnU-Net track is confirmed as the one to push. **Provenance gap, named not hidden:** EN said this in chat; no comparison artifact is pinned in the brain — which models, which cohort, which metric, which run. Before this sentence is used in a grant, a paper or a company document, the underlying comparison must be pinned to a file and a run.
Company sell-blocker #1 — "contours not good enough" — and the postproc dependency for Imaging-4.0 M1.1 & M3.1. Two-track: nnU-Net product + interactive correction; CineMA as pre-segmenter. Execution is Raphael/Luisa. Includes "how to finalize the contours" (Goethe CVI). Node: `20-company/cmr-segmentation/`.

Perfusionmed

**Set by EN 2026-08-29: *"#2 is perfusion."*** First time perfusion is ranked as a company priority rather than tracked as a person-level risk.
**It has been self-deprioritised for months.** Quang built a working prototype and EN was pushing it to a state a non-technical user can run ("quite far"), but Quang has repeatedly self-paused it for "higher-prio research/software" — it is the standing **coverage-risk** item on his PEOPLE card. Ranking it here is the counterweight to that.
**No node of its own.** It lives inside `50-team/quang/STATE.md`. If it is a company rank 2, it has outgrown that — worth its own node.
Scientific frame is settled and non-negotiable (SYSTEM.md / EN's CMR positions): **quantitative perfusion IS phenotyping** — quantitative MBF mapping is how ischaemia is phenotyped. Do not accept "phenotyping rather than quantification" framings for perfusion.

Friday block — not one of the five lanes

Idea triage — rank the open endsmed

**(EN 2026-08-29) The next Friday topic.** Run every open end through the rubric in `40-infrastructure/idea-triage/STATE.md`, rank them, and end with a read-only ledger to talk through — a discussion surface, not something he click-edits.
**Why it is here and not in the nightly queue:** it sat as `2026-06-28-idea-triage-first-pass` in `offline-jobs.md` and was re-deferred **13 times** since 2026-06-29. The deferral reason was always sound — the rubric's likelihood column has to be literature-search-grounded per SYSTEM.md §7, and that is not honest unsupervised 23:00 work. A job no sweep will ever run is stuck, not queued. Moving it here is the fix.
**Scope of "open ends":** trAIectories (Bad Orb), INCA, vp-brain-clone, trustworthy-ai-science, science-harvester, idea-triage itself, plus every initiative already in this file. Output: `_morning/idea-ledger.html` + scores to `_morning/sources/idea-triage.csv`.
**Rank deliberately left unset.** `Bad Orb` and `Cooperations` are both marked `rank: 1` in this file — a pre-existing collision, not something I resolved on his behalf. Sorting the ranks is itself a good first five minutes of this block.

Voxel-Map (MPE) — Kandidat für die leere paper-Lanehigh

**(EN 2026-08-31) „put into strategy session (high up) — this could be in the 'next paper' category, which is currently empty as all other papers require data production first."** Das ist das Unterscheidungsmerkmal: die Voxel-Map läuft auf Daten, die schon da sind. n=100 eingebettet, n=94 sauber, septal T1 1141±44 ms / T2 38,9±2,9 ms, Within-Case-CV T1 median 3,4 %; beide Werkzeuge gebaut (`mpe_voxel_explorer_n100.html`, `mpe_binned_lattice_n100.html`).
**Bewusst NICHT auf `lane: paper, rank: 1` gehoben.** EN sagte „could be", nicht „is" — und die Hausregel lautet: lieber ein zu niedriges als ein zu hohes Signal, eine Lane bleibt offen, statt dass ich einen Kandidaten befördere, den er nicht gewählt hat. Die Beförderung ist der erste Satz der Strategiesitzung.
**Namensfrage vor dem Titel klären:** EN sagt „voxel map", der Knoten sagt „mapping pixels / MPE". Zwei Namen für dieselbe Sache werden im Titel teuer.
**Ungegroundet:** ob die Voxel-Ebenen-Darstellung von T1/T2 als Phänotypisierungswerkzeug schon publiziert ist, wurde nicht recherchiert. Vor der Positionierung nachholen (SYSTEM.md §7).

In the queue (not yet discussed)

CMR-AI product roadmap (Goethe CVI) — **Candidate for the `company` lane, NOT promoted (2026-08-29).** Nothing in the brain marks a current company priority the way EN marked GREY-ZONE for grants and P3 for papers. The visible candidates are this roadmap and **Contours** (`actioned`, called *"company sell-blocker #1"*, waiting on the Luisa CorSeg verdict + Raphael's nnU-Net baseline). Promoting one would be me choosing; the lane stays open until EN says which.
Branding — Eike + VP personal brand — uni prof + scientist + high-end practice + healthy-lifespan, AI-supported, evidence-based. Peter-Attia-style with chair-level research credentials. The brand is the people. Overlaps Bad Orb.
Task-list / brain-system cleanup — **(Eike 2026-06-18, via feedback)** "Clean up the whole task list." The flat TODO list has accreted project-status items that aren't actionable todos. Target shape: keep only the 3–5 todos a working session actually produces; everything else is either project status (lives in STATE.md, not the flat list) or a parked/triggered item. Likely touches `sweep_todos.py` bucketing + what the overview/task-console surface by default.
Research roadmap ↔ strategy overview (merge the two surfaces) — **(2026-08-18, Eike):** the Gantt built today (`10-research/research-roadmap_grants-and-AI-papers_2026-08-18.html`) and this strategy overview are "the same thing in a different format". Merge them into one surface that shows, per parallel area, **what the next focus should be while the longer-term projects keep moving forward**. Explicitly NOT now — parked until raised.
Reimbursement reconciliation ledger (Reisekosten + Barmenia) — **(Eike 2026-07-27, parked — "full to the rim today, bring up on a quieter day")** The recurring leak: no ledger tracking *spent → eingereicht → erstattet* per claim. Previous secretary → several 1000s € lost. Two gaps: paid-but-never-submitted (pure loss) and submitted-but-rejected. ⛔ decide scope (travel only vs + Barmenia) + where the "spent" column comes from (card/bank statements) — without it the tool only catches submitted-vs-reimbursed, not never-submitted (the bigger leak).

Show actioned — no intervention until trigger (6)

Full Imaging Projectuntil: Gantt needs a review

Milestones/timeframes/Gantt worked 2026-06-05; execution now flows via delegations + action points, not strategy.

MR-vendor strategyuntil: **unbefristet geparkt (EN 2026-08-31: „park")** — der alte Trigger `2026-07-09` war 53 Tage abgelaufen und von niemandem eingelöst worden. Jetzt bewusst ohne Datum statt scheinbar terminiert.

Get a real alternative / second source away from the majors. Lens: field strength, cost, CMR depth, openness to external tools (own feedback loop + AI-supported scanning). Narrow eval to Canon Vantage Galan 3T + United Imaging 5T vs Siemens/GE baseline — see § MR vendor landscape below.
Includes SpinLab hyperpolarizer feasibility on the FFM Prisma (X-nucleus channel + dual-tuned coil + 13C sequences; SPINlab capital ~€1.5–2.5M; QP/pharmacist for human IMP release). Park as real eval if the Würzburg 22-Jun discussion points that way.
**(2026-06-07)** Keep the full vendor table; ALSO prepare a SHORT 3T + low-field version with a table of the key questions to start asking vendors. Back burner until the DFG grant is submitted, then activate vendor contact.
**(2026-06-11) Eike: low flame — after Grants. Park 4 weeks, do NOT show in the list, then bring back (resurface 2026-07-09).** (Resolves the unclear 06-09 note: this initiative = Eike's list-item #4.)

DZHK Professors — spokesperson roleuntil: after the Eike·Tanja·Ralf kick-off call (resurface to confirm slate + board pitch)

**Worked through 2026-06-09 (strategy session closed).** Eike elected speaker 2026-06-01 (term Jun 2026 – May 2028; Tanja Zeller outgoing, Ralf Gilsbach incoming). Strategy: two engines — (A) board-mandated tasks, Eike as router not labourer; (B) uninvited "DZHK view on X" position papers, mixed authorship.
Slate to the board: **AI** (Eike leads — flagship, "get invited"), sex & gender, sustainability, + staged healthy-longevity (round 2). Signature instrument = the connection map → offer the board a **portfolio map of the DZHK project database** (links / clusters / parallel-work / opportunities / holes); demo-first (public SE network) then ask for the DB.
Full record + pre-call brief → `10-research/admin/research-politics/dzhk-professors/STATE.md` + `outputs/2026-06-09_pre-call-brief_triumvirate.md`.

DZHK HFpEF focus-area (Focus group)until: 2026-07-15

Focus-area co-lead governance + cluster map. Scaffolding at 10-research/admin/research-politics/dzhk-hfpef-focus-area/.
**(2026-07-03) Eike: no strategy from us for the moment.** A retreat catch-up (Gabriele Schiattarella + Sandy Engelhardt + Eike) is set for **14 Jul** — do NOT bring up the retreat before then. **After the 14th, remind Eike to spend 30 min thinking about a HFpEF strategy** (resurface trigger = 2026-07-15).

trAIectoriesuntil: **2026-09-21** (EN 2026-08-31: „own vehicle, but can rest a bit — remind me in 3 weeks, after DZHK meeting")

**Die Vehikelfrage ist entschieden: eigenes Vehikel.** Damit fällt der alte Trigger weg, der auf die Bad-Orb-/Mojácar-Struktur wartete — eine Kette aus zwei Wartenden, von denen keiner sich bewegte. Jetzt ein Datum statt einer Bedingung.
New company? The persistent software layer (clinical-case-UI / trAIectory app) that runs on top of both Bad Orb and Mojácar — decide whether it's its own vehicle.
**(2026-06-11) Eike: decided — the Trajectories branding/naming was worked through 2026-06-09 / Tue; positioning locked (pending VP weekend sign-off). Folds into Branding execution; off the strategy queue for now.**

Longevity / Mojácar (long-term)until: Eike reactivates (he will bring it up actively)

The destination longevity centre (Spain) that Bad Orb funds the path to. Long-horizon; the eventual flagship.
**(2026-06-11) Eike: very low flame — park for now; he'll bring it up actively when ready.**

Add by dropping @strategy <topic> into any STATE.md, talkback, or tell Claude. Ranks, deadlines & states live in _morning/STRATEGY.md (Claude-maintained).

Note about Strategy (autosaves; send when ready)

brain — overview

Inbox — needs your attention

Due today

Today's calendar — Friday, 2 October 2026

Deadlines

Open decisions across the brain

Work next — one per lane

Next grant

Next paper

Next collaboration

Next private-praxis endeavour

Next company priority

Friday block — not one of the five lanes

In the queue (not yet discussed)