Nova Gov — Hospital Intelligence Platform (Demo)

Bed Occupancy

93%

230 / 247 beds occupied

DOH threshold: 85%

ER Avg Wait

28 min

↓ 11 min vs last week

Target: <45 min (DOH EB)

Active Admissions

18

↑ 3 since 06:00

8 ER, 7 OPD, 3 transfers

PhilHealth Claims

14

pending >72h · action needed

₱2.1M at risk of rejection

Discharges Today

11

↑ 2 vs yesterday

3 pending clearance

✨ Nova Gov · Hospital Intelligence — what it's doing right now

AI-powered hospital operations for private and government hospitals in the Philippines

₱2.1M claims at risk — action needed

⚠️

Bed capacity alert: ICU at 100%, general wards at 96%

At current admission rate, general wards reach 100% by 14:30. AI recommends activating surge protocol and contacting District Hospital Lipa City for 6 available beds. Transfer authorization pre-drafted.

surge alert

💳

Auto-filed 23 PhilHealth claims this morning — ₱1.8M submitted

AI extracted diagnosis codes, matched to case rates, validated against PHIC guidelines, and submitted electronically. Manual process typically takes 2–3 days per billing staff. Done in 4 minutes.

₱1.8M filed

🚨

14 claims approaching 72h deadline — ₱2.1M at risk of rejection

Missing: 6 physician signatures, 4 operating room logs, 4 lab attachments. AI has drafted completion checklists per claim and auto-notified responsible departments. Requires human sign-off only.

₱2.1M at risk

🔮

Predicted: 22 admissions expected today based on ER volume + day-of-week pattern

Monday patterns + barangay health referrals + weather (rain, Batangas) suggest above-average ER load. Staffing adjustment recommended: +2 ER nurses 13:00–21:00. Pre-notified nursing supervisor.

staffing ready

💊

Pharmacy: 3 critical stock items below 2-day buffer — auto-reorder initiated

Amoxicillin 500mg, IV fluid NaCl 0.9% 1L, and Paracetamol 500mg tabs flagged. PhilGEPS-aware procurement check passed. Purchase orders drafted for medical officer approval.

reorder ready

Recent admissions

Last 8 patients · live from admissions desk

View ER queue →

Patient ID	Chief Complaint	Priority	Ward	Status	PhilHealth
PT-20260519-A	Chest pain	STAT	ICU Bed 3	Critical	Pending
PT-20260519-B	Dengue fever	Routine	Ward 2 Bed 14	Admitted	Filed
PT-20260519-C	Fracture (RTA)	STAT	Ortho Ward	Preop	Case rate
PT-20260519-D	Pneumonia	Routine	Medical Ward	Admitted	Filed
PT-20260519-E	GI bleed	STAT	ICU Bed 7	Critical	Review
PT-20260519-F	Hypertensive crisis	Routine	Medical Ward	Stable	Filed
PT-20260519-G	OB — active labor	STAT	DR Suite	In progress	Case rate
PT-20260519-H	Post-op monitoring	Routine	PACU Bed 2	Monitoring	Filed

Ward census

Bed occupancy by department

8/8

ICU (100%)

48/50

Medical Ward

32/36

Surgical Ward

22/24

OB-Gyne

18/24

Pediatrics

40/50

Ortho / Rehab

28/30

ER Obs Beds

34/25

Private Rooms

Surge protocol threshold at 95% overall. Current: 93%. AI projects breach by 14:30 at current admission rate.

Department status

Live from hospital information system · updated every 2 min

View all →

ER

Emergency Room

Waiting8

ICU

Intensive Care

Occupancy100%

OR Suite

3 theaters

Active2 / 3

Pharmacy

Central dispensing

Low stock3

Radiology

X-ray · CT · Ultrasound

Queue12

Active alerts

Patient safety · bed capacity · claims · compliance

Acknowledge all →

🚨

Critical: ICU at 100% — no beds available for new critical admissions

Auto-notified CMO Dr. A. Santos · 09:41 AM · Nearest available: District Hospital Lipa (6 ICU beds) · transfer pre-authorized

💳

14 PhilHealth claims approaching 72h filing deadline — ₱2.1M at risk

Missing: physician signatures (6), OR logs (4), lab attachments (4) · checklists sent to departments · 09:38 AM

⚠️

Surge warning: bed occupancy projected to reach 95% by 14:30

Based on Monday admission pattern + current ER queue of 8 · surge protocol draft ready for CMO sign-off

💊

Pharmacy low stock: Amoxicillin 500mg, NaCl 0.9% 1L, Paracetamol tabs

Auto-reorder PO drafted · awaiting medical officer approval · 09:14 AM

📋

DOH bed capacity report due end of week — 87% complete

AI pre-filled from HIS data · requires infection control sign-off · no manual data entry needed

Sample ID	Patient	Type	Priority	Instrument	Operator	Received	TAT	Status	Flags
H-26051509-148	PT-44781	CBC + Diff	STAT	XN-1 (Hema-A)	R. Mendoza	09:24 AM	17 min	Critical	Blasts?WBC ↑↑
H-26051509-147	PT-19402	CBC + Diff	Routine	XN-2 (Hema-B)	R. Mendoza	09:18 AM	38 min	Verified	NRBC#
H-26051509-146	PT-30817	CBC + Diff + Retic	Routine	XN-1 (Hema-A)	J. Cruz	09:11 AM	44 min	Verified	Aniso/Poikilo
H-26051509-145	PT-22059	CBC + Diff	STAT	XN-2 (Hema-B)	J. Cruz	09:08 AM	32 min	Running	—
H-26051509-144	PT-66104	Retic count	Add-on	XN-3 (Hema-C)	A. Santos	09:02 AM	49 min	Verified	—
H-26051509-143	PT-71289	CBC + Diff	Routine	XN-1 (Hema-A)	R. Mendoza	08:57 AM	54 min	Verified	Atyp. Lympho?
H-26051509-142	PT-54021	CBC + Diff	Routine	XN-2 (Hema-B)	A. Santos	08:51 AM	1h 02 m	Hold · QC	Plt Clumps?
H-26051509-141	PT-49832	CBC + Diff	Routine	XN-3 (Hema-C)	A. Santos	08:48 AM	41 min	Verified	—
H-26051509-140	PT-30994	CBC + Diff	STAT	XN-1 (Hema-A)	R. Mendoza	08:42 AM	21 min	Critical	Hgb ↓↓Microcytic
H-26051509-139	PT-58219	CBC + Diff	STAT	XN-2 (Hema-B)	J. Cruz	08:30 AM	24 min	Critical	Plt ↓↓
H-26051509-138	PT-12087	CBC + Diff + ESR	Routine	XN-3 (Hema-C)	A. Santos	08:24 AM	47 min	Verified	Hypochromic
H-26051509-137	PT-90215	CBC + Diff	Routine	XN-1 (Hema-A)	R. Mendoza	08:18 AM	36 min	Verified	—
H-26051509-136	PT-77450	CBC + Diff	Routine	XN-2 (Hema-B)	R. Mendoza	08:11 AM	42 min	Verified	IG #
H-26051509-135	PT-66821	CBC + Diff	Routine	XN-3 (Hema-C)	J. Cruz	08:04 AM	39 min	Verified	—
H-26051509-134	PT-41208	CBC + Diff + Retic	Add-on	Retic-1	A. Santos	07:58 AM	52 min	Verified	MacrocyticRET-He ↓

Showing: last 30 runs · Bio-Rad Liquichek tri-level

WBC · Level 1 (Low)

Target: 3.20 ×10⁹/L · SD: 0.16 · CV: 5.0%

PASS

Last run · 3.18 30-day CV · 4.6% Outliers · 0

MCV · Level 2 (Normal) Hema-B

Target: 86.0 fL · SD: 1.2 · CV: 1.4%

WARN · 10x trend

Last run · 88.1 30-day CV · 1.8% Outliers · 2

HGB · Level 2 (Normal)

Target: 13.2 g/dL · SD: 0.20 · CV: 1.5%

PASS

Last run · 13.24 30-day CV · 1.3% Outliers · 0

PLT · Level 2 (Normal)

Target: 256 ×10⁹/L · SD: 12 · CV: 4.7%

PASS

Last run · 258 30-day CV · 4.2% Outliers · 0

Westgard rule status · last 30 runs

Multi-rule QC · auto-evaluated per parameter

1₃ₛ

One control point outside ±3SD

PASS

2₂ₛ

Two consecutive points outside ±2SD same side

PASS

R₄ₛ

Two consecutive points with a 4SD range

PASS

4₁ₛ

Four consecutive points outside ±1SD same side

PASS

10ₓ

Ten consecutive points on the same side of the mean — Hema-B MCV L2

REVIEW

AI corrective-action suggestion

Generated from the Hema-B MCV trend

Trend signal: 10x rule borderline on MCV Level 2 (Hema-B) 10 consecutive points have drifted above the mean (+1.5 to +1.8 SD). No 1_3s or 2_2s failures yet, but the drift pattern matches a typical aperture-clog or diluent-line bias.

Recommended actions, in order:

Run cleaning cycle on Hema-B RBC channel (CELLCLEAN auto). ETA 6 min.
Re-run Level 2 control × 2.
If still drifted: recalibrate MCV against fresh primary calibrator (CAL-Set).
If post-cal CV stays > 1.6%, escalate to service. Suspected: HGB photometer drift.

Sysmex equivalent: SNCS Insight remote-call. Nova flags this 28 hours earlier than vendor-side telemetry typically catches it.

Case 1 of 2 · Critical · AML pattern · click any sample row to switch · two example cases loaded for demo

H-26051509-148 · PT-44781

Reyes, Maria L.

42 F · Hema-Onco clinic · MR# H-44781 · DOB 1983-08-14

STAT Critical · MD called 5-part Diff + Retic

Received09:24 AM

Run09:38 AM · XN-1

TAT17 min

OperatorR. Mendoza, MT

Complete Blood Count

Parameter	Result	Flag	Reference
WBC ×10⁹/L	168.3 ↑↑	Critical High	4.0 – 10.0
RBC ×10¹²/L	3.21 ↓	Low	4.10 – 5.30
HGB g/dL	9.4 ↓	Low	12.0 – 16.0
HCT %	28.6 ↓	Low	36.0 – 46.0
MCV fL	89.1	—	80.0 – 100.0
MCH pg	29.3	—	27.0 – 33.0
MCHC g/dL	32.9	—	32.0 – 36.0
RDW %	17.8 ↑	High	11.5 – 14.5
PLT ×10⁹/L	28 ↓↓	Critical Low	150 – 400
MPV fL	11.4 ↑	High	7.5 – 10.4

5-Part Differential

Parameter	%	Abs (×10⁹/L)	Reference
Neut	14.2 ↓	23.9	50 – 70 %
Lymph	12.1	20.4 ↑	20 – 40 %
Mono	4.8	8.1	2 – 8 %
Eos	0.9	1.5	1 – 4 %
Baso	0.4	0.7	0 – 1 %
Blast?	67.6 ↑↑	113.8	Not detected (normal)

Nova AI interpretation

Acute myeloid leukemia pattern — high confidence

Hyperleukocytosis (WBC 168.3 ×10⁹/L) with profound thrombocytopenia (28 ×10⁹/L), normocytic anemia (HGB 9.4 g/dL), and a 67.6% blast population on the auto-diff with prominent left-shift and absent maturing granulocytes. Pattern is most consistent with acute leukemia. Manual smear required for blast confirmation and morphologic typing (myeloid vs lymphoid).

Acute Myeloid Leukemia 74%

Acute Lymphoblastic Leukemia 18%

Severe sepsis with left-shift 6%

Other (CML blast crisis, etc.) 2%

Recommended next steps (auto-triggered):

Smear queued for manual differential (priority STAT)
Flow cytometry tube auto-reserved · MAPI panel
Dr. Reyes (Hema-Onco on-call) paged · acknowledged 09:41 AM
Coag panel + LDH + uric acid suggested (tumor lysis risk)

Sysmex equivalent: CAM flags "Blasts?" — provides no differential pattern matching, no next-step orchestration. Nova auto-ordered 3 reflexes in 8 seconds.

Patient delta · WBC trend

PT-44781 · last 6 visits over 18 months

Δ WBC: 6.8 → 168.3 ×10⁹/L · 24.7× jump from her March '26 baseline. Nova auto-detects delta-check breaches across all longitudinal patient data.

Smear preview · queued for review

Auto-captured at flagging · 40× / 100×

40× Field 1

40× Field 2

100× Oil

AI-pre-counted 62% blast forms, scattered Auer rod–like structures noted. Manual confirmation by pathologist required.

Audit trail

Tamper-evident · ASTM E1394 compatible

09:24 AM · Sample received (R. Mendoza)

09:32 AM · Loaded on Hema-A (XN-1)

09:38 AM · Results released by analyzer

09:38 AM · Nova AI: critical pattern flag

09:38 AM · Auto-page Dr. Reyes (sent SMS + Slack)

09:41 AM · Dr. Reyes acknowledged

09:41 AM · Reflex tests auto-ordered (Flow, Coag, LDH, UA)

09:42 AM · Smear queued for manual review

Awaiting manual smear verification before final release.

H-26051509-146 · PT-30817

Ramos, Joel S.

28 M · Outpatient annual physical · MR# H-30817 · DOB 1997-03-22

Routine Verified · auto-released AI insight available

Received09:11 AM

Run09:51 AM · XN-1

TAT44 min

OperatorJ. Cruz, MT

Complete Blood Count

Parameter	Result	Flag	Reference
WBC ×10⁹/L	6.8	—	4.0 – 10.0
RBC ×10¹²/L	5.42 ↑	High · key clue	4.10 – 5.30
HGB g/dL	11.2 ↓	Low	12.0 – 16.0
HCT %	34.1 ↓	Low	36.0 – 46.0
MCV fL	67.4 ↓↓	Microcytic	80.0 – 100.0
MCH pg	20.7 ↓	Hypochromic	27.0 – 33.0
MCHC g/dL	30.7 ↓	Low	32.0 – 36.0
RDW %	16.8 ↑	High · dual pathology clue	11.5 – 14.5
PLT ×10⁹/L	312	—	150 – 400
MPV fL	9.2	—	7.5 – 10.4

5-Part Differential

Parameter	%	Abs (×10⁹/L)	Reference
Neut	56.2	3.82	50 – 70 %
Lymph	34.1	2.32	20 – 40 %
Mono	6.4	0.43	2 – 8 %
Eos	2.8	0.19	1 – 4 %
Baso	0.5	0.04	0 – 1 %

Computed indices

Index	Value	Interpretation	Decision rule
Mentzer index (MCV / RBC)	12.4	Thal-trait pattern	< 13 → thal · > 13 → IDA
Green & King index	63.8	Thal-trait pattern	< 65 → thal-trait
RDW Index (RDW × MCV / RBC)	209	Borderline	> 220 → IDA likely

Nova AI interpretation

Beta-thalassemia trait with emerging iron deficiency overlay

The pattern is classic for thal-trait: microcytic + hypochromic indices with a high-normal RBC count (5.42 ×10¹²/L) and Mentzer index 12.4. A pure iron-deficiency anemia would show a low RBC count and Mentzer > 13. The elevated RDW (16.8%) is the dual-pathology signal — pure thal-trait keeps RDW ≤ 14.5. Over 12 months this patient's MCV has drifted from 78 → 67.4 fL while RBC stayed stable, consistent with emerging iron deficiency layered on a thal-trait background.

Beta-thalassemia trait (existing) 81%

+ Iron-deficiency overlay (new) 68%

Alpha-thalassemia trait (alt) 14%

Anemia of chronic disease 5%

Suggested reflexes (pending physician approval — no auto-order on routine):

Hb electrophoresis (HbA₂ & HbF) — confirm thal-trait
Ferritin, serum iron, TIBC, transferrin saturation — confirm IDA layer
Reticulocyte hemoglobin (RET-He) — earliest iron-restricted erythropoiesis marker

Sysmex equivalent: CAM flags "Microcytic anemia" without distinguishing thal vs IDA or detecting the 12-month delta. The Mentzer/G&K math + the longitudinal pattern is the Nova-only catch.

Patient delta · MCV trend

PT-30817 · 5 visits over 12 months

Δ MCV: 78.0 → 67.4 fL over 12 months, while RBC count stayed flat at 5.3–5.5. The slow drift + stable RBC = iron stores depleting on a thal-trait background. No single visit would have triggered a flag; Nova caught it across the longitudinal series.

RBC volume distribution

Histogram · target 80–100 fL · this patient: shifted-left + bimodal

Bimodal microcytic distribution + small "shoulder" near normal range. AI uses raw histogram (not just MCV) to separate pure thal-trait (sharp unimodal < 80) from thal + IDA overlay (bimodal). Sysmex software shows the histogram but does not classify the pattern.

Workflow context

Why this catch matters

Volume bucket: Routine adult male CBC. Today: 184 of 348 samples (53%).

Manual review risk: Without AI, this would have been auto-released as "microcytic, NOS" by a tech under time pressure. No reflex would have been ordered.

Clinical cost of miss: Patient continues to drift toward overt anemia. Caught at year 5–7 typically, in the PH primary-care setting.

Nova caught this on a routine sample. Suggested reflexes were surfaced to the MD inbox without blocking the result release. Patient is now on the workup queue.

Auto-route + escalation enabled · MD callback SLA: 5 min

Critical right now

3

2 unacknowledged

Auto-paged on-call MD

Avg time to callback

4m 18s

↓ 38% vs last quarter

Phone + SMS confirmed

Last 24 h

12

8 acknowledged, 3 escalated, 1 open

Trending +2 vs 30-day avg

Unack > 15 min

0

↓ from 1.4 / day (Q1)

Director escalation triggers

Active critical values

Click a row to see the AI-routed callback chain

Patient	Parameter · Value	MD on-call	TTC	Status
PT-44781 Reyes, Maria L. · 42 F	WBC 168.3 ×10⁹/L + blasts on auto-diff	Dr. Reyes (Hema-Onco)	3m 12s	Paged · awaiting callback
PT-30994 Cruz, Reyna M. · 56 F	Hgb 4.2 g/dL Microcytic, suspect GI bleed	Dr. Tan (ER)	9m 41s	Paged · ER notified
PT-58219 Domingo, Reyna · 38 F	Platelet 8 ×10⁹/L Post-induction chemo	Dr. Reyes (Hema-Onco)	—	Ack 08:52 · Plt ordered
PT-29071 Sy, Rachelle · 29 F	K⁺ 6.4 mmol/L Hyperkalemia, AKI workup	Dr. Garcia (Neph)	—	Ack 07:14 · IV insulin started
PT-77450 Lim, Mark · 64 M	Glu 28 mg/dL Severe hypoglycemia, ICU	Dr. Bautista (ICU)	—	Ack 06:48 · D50 given
PT-12087 Tan, Robert · 71 M	Troponin I 38.4 ng/mL STEMI confirmed	Dr. Mendoza (Cardio)	—	Ack 06:11 · Cath lab

AI-routed callback chain · PT-44781

Auto-paged Hema-Onco on-call · escalation chain armed

PAGED 09:38 · 3m 12s

09:38 AM SMS sent · Dr. Reyes (+63 917 ··· 442) · "Critical WBC 168.3 ×10⁹/L on PT-44781. Blasts on auto-diff. Smear queued. Reply 1 to ack."

09:38 AM Slack DM · @joyce.reyes · same payload · clickable result link · marked unread

09:41 AM No callback yet · ~ 2 min to backup-MD escalation · backup is Dr. Bautista · auto-page SMS scheduled at 09:43

09:43 AM Backup MD page (Dr. Bautista) — pending

09:53 AM Director escalation (Dr. Aquino) — pending

AI-suggested follow-on orders

Peripheral smear review

Auto-queued at smear station

Flow cytometry panel

Tech approval pending

Coag panel + D-dimer

DIC workup on auto-diff blasts

Blood bank type & cross

2 units PRBC + 1 unit Plt

Sysmex equivalent: IPU surfaces the critical value as a flagged result. It does not page on-call, does not escalate, does not orchestrate downstream orders. Manual phone tag is the default.

Sorted by AI urgency × uncertainty · highest-impact, lowest-confidence first

Flagged for smear review

AI labels its own confidence — sub-60% cases bubble up

Patient	Primary flag	AI top dx	Confidence
PT-44781 Reyes, Maria L.	Blasts?	AML, possibly M2	54%
PT-71289 Domingo, Reyna	Atyp lympho?	Reactive lympho · viral	48%
PT-54021 Sy, Rachelle	Plt clumps?	EDTA-PTCP (false low)	84%
PT-77450 Tan, Mark	IG # ↑	Left shift · sepsis workup	78%
PT-58219 Lim, Mark	Schistocytes?	TTP vs DIC · ambiguous	41%
PT-49832 Garcia, Luis	NRBC #	Marrow stress · benign	88%
PT-66821 Bautista, Joel	Macrocyte	B12 / folate deficient	72%

PT-44781 · smear preview + AI breakdown

3 microscope fields · automated cell-level segmentation

CONFIDENCE 54%

Field 1 · 100× oil

Field 2 · 100× oil

Field 3 · 50× scan

AI ranked differential:

54% AML, FAB M2 (granulocytic blasts)

22% AML, FAB M4 / M5 (monocytic)

14% ALL (lymphoid blasts)

10% Reactive lymphocytosis (rule-out)

Why the AI is uncertain: classifying M2 vs M4 needs Auer rods + cytochemistry. The blast cytoplasm shows granules but they are not consistently cell-by-cell. Recommend pathologist review with myeloperoxidase stain before final dx.

Sysmex CAM equivalent: auto-releases or holds the sample. Does not rank the hold queue by risk-weighted urgency, does not label its own uncertainty, does not differentiate blast subtypes. Techs triage by hand.

5 sites · 1,847 samples across the network today

⚠️

Quezon City running 1.7× network-median TAT — 14 STAT samples backed up

Makati capacity available · Nova AIOS recommends courier rerouting · approve to dispatch

BGC · Taguig

412

samples today

TAT38 m

QC pass99.1%

Critical1

Makati

348

samples today

TAT42 m

QC pass98.7%

Critical3

Quezon City · Δ

298

samples today · 14 backlog

TAT71 m

QC pass97.4%

Critical2

Cebu

389

samples today

TAT44 m

QC pass99.3%

Critical0

Davao

400

samples today

TAT41 m

QC pass98.9%

Critical1

✨ AI load-balancing recommendation

Cross-site dispatch model · approve to execute

CONFIDENCE 91%

Reroute 6 STAT samples from Quezon City → Makati via courier dispatch at 11:30 AM.

QC instrument signature matches between the two sites within ±0.4 SD on all 8 control lots. Makati has 3 of 4 analyzers idle for the next 45 minutes. Courier transit time is 38 minutes door-to-door at this hour. Net reduction in patient TAT: 34 minutes per sample.

Sysmex Caresphere equivalent: shows multi-site dashboards as a paid add-on. Does not auto-recommend rerouting. Does not check QC-signature compatibility before dispatch. Manual decision.

Cross-site QC drift · MCV L2

Same Bio-Rad lot across all 5 sites · last 14 runs

QC · 1 site drifting

BGC Makati Quezon City (drifting) Cebu Davao

4 of 5 sites tracking on lot mean within ±0.5 SD. Quezon City has drifted +1.8 SD over the last 8 runs — same pattern as the Hema-B drift surfaced on the QC Trends screen. Single-site issue, not lot-related. Nova default includes this comparison; Sysmex requires Insight per-site SaaS.

H-26051509-146 · PT-30817

Ramos, Joel S.

28 M · Routine outpatient annual physical · CBC + Diff

Total chain44 min

Touched by3 humans · 2 AI agents

Audit refs17 events

ComplianceISO 15189 ✓

Sample chain of custody

Every event signed, timestamped, and replayable. AI decisions logged with model + prompt context.

Open full audit trail →

Sample registered

Order #A-26051509-001 from OPD-Internal-Med, Dr. Aquino. EDTA tube, 3.0 mL. Barcode scanned at receiving.

Operator: J. Cruz, MT (MTAS-2018-4421) · Workstation: REC-02

09:11 AM

Aliquoted + routed

Volume check pass (3.1 mL). Routed to Hema-A per active method HEM-CBC-DIFF-v4.2. Pre-analytical hold: none.

Auto-routed by Nova AIOS sample-router · matched method on order type + sample type + patient age band

09:14 AM

Analyzer run started

Hema-A (XN-1000, S/N 27814). Reagent lot SL-25-118. QC level 2 ran 28 min prior (PASS, MCV 86.3).

Method version locked: HEM-CBC-DIFF-v4.2 · revision approved 2026-02-10 by Dr. Reyes

09:42 AM

AI pattern interpretation

Model: nova-lab-hema-1.2.3 · Confidence: 81% · Pattern: β-thal trait + emerging IDA overlay. Mentzer 12.4. RDW 16.8. 12-month MCV drift detected. Suggested reflexes: Hb electrophoresis, ferritin, RET-He.

Decision logged · prompt + response stored under audit-log #A-26051509-0817 · model retention: 7 years

09:51 AM

AI auto-validation

All 18 reportable values within method limits. No delta-check failures vs prior visits. Result auto-released per active SOP VAL-CBC-AUTO-v3.0. AI confidence: 99.4%.

Auto-release rule: PT-30817 not flagged in critical-watch list · no manual override required

09:55 AM

Result published

Pushed to LIS via HL7 ORU^R01. Patient portal updated. Dr. Aquino's MD-inbox received the result + AI's suggested reflexes (not auto-ordered on routine outpatient).

Verified by: R. Mendoza, MT (Chief MT, signature 0x8e21) · electronic signature stored per 21 CFR Part 11 §11.50

09:55 AM

Stability retention (planned)

Sample retained at 2-8°C in cold-storage rack CS-A-03-12. Discard date: 2026-05-22 09:55 AM (7-day retention per ISO 15189 §5.7).

Nova AIOS will auto-discard + log when the timer expires · no human action required unless add-on test ordered

in 7 days

Archive event (planned)

Sample physical discard + lifecycle finalize. Result data retained 25 years per DOH-PH retention rules. AI prompt + response retained 7 years.

Auto-fires at 09:55 AM, 2026-05-22 · audit-log entry will be #A-26051509-0817-final

2026-05-22

Sysmex IPU equivalent Shows result status. Sample chain stops at the analyzer. No record of who reviewed, which method was active, or how AI reached its conclusion. ISO 15189 chain-of-custody is left to the LIS.

OpreX LIMS equivalent Strong sample lifecycle and 21 CFR Part 11 logging. AI decisions are out of scope; Yokogawa does not log the model + prompt that produced an automated release. Nova does, with 7-year retention.

ISO 15189 §5.5.1 · 14 active SOPs · 1 in revision

Active methods + SOPs

Version-locked. Operators must be certified on the active version before they can run a sample.

+ New method draft

Method	Discipline	Version	Status	Certified
HEM-CBC-DIFF 5-part differential CBC	Hematology	v4.2	Active	14 of 16 MTs
HEM-RETIC Reticulocyte count	Hematology	v2.1	Active	11 of 16 MTs
HEM-BODY-FLUID CSF / body fluid count	Hematology	v1.4	Active	6 of 16 MTs · 2 due recert
HEM-SMEAR-REV Manual peripheral smear review	Hematology	v3.0	Active	9 of 16 MTs + 2 pathologists
COA-PT-INR Prothrombin time + INR	Coagulation	v2.5	Active	7 of 16 MTs
COA-APTT Activated PTT	Coagulation	v2.4	In revision	— pending re-cert
VAL-CBC-AUTO CBC auto-release decision rule	Validation	v3.0	Active	All MTs (auto)
VAL-CRITICAL-PAGE Critical-value paging chain	Validation	v2.2	Active	All MTs (auto)

⚑ AI-detected method drift

Live behavioral analysis vs the validated method spec

REVIEW

HEM-CBC-DIFF v4.2 · 3 operators outside spec Over the last 14 days, 3 of 14 certified MTs have been running samples through Hema-A with average specimen-at-analyzer temperature +2.1°C above the method's validated range (18–22°C). Method drift, not analyzer drift. AI surfaced this from the per-sample environmental telemetry.

Recommended actions:

Re-train 3 named MTs on HEM-CBC-DIFF v4.2 §6.3 (specimen handling).
Calibrate the receiving-station thermometer (last cal: 8 months ago — overdue).
Either re-validate the method up to 24°C or post a visible reminder at REC-02.

Sysmex / OpreX equivalent: store the SOP. Show who is certified. Neither vendor monitors method drift from telemetry. Nova catches it because every sample is a data point against the method's validated parameters.

Revision pipeline + e-signature stack

21 CFR Part 11 §11.10 (b)(e) — every method change captured with reason, reviewer, timestamp

In revision

COA-APTT v2.4 → v2.5

Tightening upper reference limit from 38s to 35s per CAP 2026 guidance. Draft by Dr. Reyes, awaiting Dr. Aquino countersign.

Just approved

VAL-CBC-AUTO v3.0

Auto-release threshold lifted from 88% to 95% of CBCs after 6 months of 99.1% concordance vs manual review.

Scheduled review

HEM-SMEAR-REV v3.0

Annual review window opens 2026-06-12. Will incorporate Nova AI confidence-uncertainty labelling per the Smear Review screen.

Audit log · last 24 hours

Every action signed, hashed, timestamped. AI decisions log the model + prompt + response. WORM-storage, 25-year retention.

Time	Actor	Action	Target
09:55:14	R. MendozaMT, sig 0x8e21	E-signature	H-26051509-146 · result release
09:55:08	nova-lab-hema-1.2.3model · auto	AI auto-release	H-26051509-146 · 99.4% conf · VAL-CBC-AUTO-v3.0
09:51:42	nova-lab-hema-1.2.3model · auto	AI interpretation	H-26051509-146 · β-thal trait + IDA · 81%
09:43:18	Dr. J. ReyesMD, sig 0x4a17	Override	H-26051509-142 · Plt 32 → 28, reason "EDTA clump verified on smear"
09:38:02	nova-lab-routing-0.9model · auto	AI critical-page	PT-44781 · WBC 168.3 ↑↑ + blasts? · Dr. Reyes paged
09:30:05	Systemcron · QC bundle	Create	QC run record · Bio-Rad Liquichek L2 · all parameters within ±2 SD
09:14:55	Dr. J. ReyesMD, sig 0x4a17	Method update	VAL-CBC-AUTO v2.9 → v3.0 · auto-release threshold 88% → 95%
09:11:22	J. CruzMT, sig 0x1c44	Sample register	H-26051509-146 · Ramos, Joel S. · OPD-Internal-Med
09:08:11	nova-lab-router-0.9model · auto	AI routing	H-26051509-145 (STAT) · Hema-B selected · load + method match

AI decision forensics · 09:51:42

Click any AI row → see the model, prompt, raw response, and confidence

Decision IDA-26051509-0817 Modelnova-lab-hema-1.2.3 · cerebras-qwen-3-235b backend Contextlab-hematology-patient Tokensin: 287 · out: 412 · latency 1.4 s Triggered byH-26051509-146 release event

Input snapshot

CBC: {WBC:6.8, RBC:5.42, HGB:11.2, MCV:67.4, MCH:20.7, MCHC:30.7, RDW:16.8, PLT:312}
Indices: {mentzer:12.4, green_king:63.8, rdw_index:209}
History: 12-month MCV 78.0 → 67.4, RBC stable

AI decision

β-thal trait + emerging IDA overlay · 81% confidence

Suggested reflexes: Hb electrophoresis, ferritin / TIBC, RET-He. Did NOT auto-order — routine outpatient, MD approval required per VAL-CBC-AUTO §4.2.

Why this is the regulator-facing differentiator: when an AI auto-released or auto-flagged a result, a DOH or CAP auditor can ask "which model? trained on what? what was the input?" and get the answer in one click. Sysmex CAM and OpreX rule engines treat their decisions as opaque. Nova logs them.