Clinical Methodology and References

The clinical combinations used in HemaLens are built upon explainable rule sets and international reference sources. This page summarizes how the system evaluates laboratory data, the clinical logic it relies on, and how it generates its outputs.

Important Notice HemaLens does not diagnose or recommend treatment. The sources presented on this page are provided to explain the reference framework that constitutes the system's clinical evaluation logic. System outputs do not replace clinical verification; the final clinical decision belongs entirely to the physician.

How the System Works

1

Parameter Evaluation

82 laboratory parameters are classified according to threshold values defined in international guidelines. Each threshold is traceably defined to a specific source.

2

Combination Analysis

Some parameters that appear normal individually may point to a different clinical picture when evaluated together. 135 literature-defined combinations are evaluated simultaneously.

3

Trend Analysis

Changes over time are made visible by comparing with the same patient's previous measurements. Changes exceeding defined thresholds are flagged.

4

Temporal Trajectory Classification

Multi-parameter trends are classified across 10 clinical groups (Inflammation, Erythrocyte, Platelet, Biochemistry, Liver, Kidney, Coagulation, Metabolic, Sepsis, Hemolysis) into 4 top-level clinical trajectories: Deterioration, Improvement, Acceleration, and Spreading. These trajectories multiplicatively modify the base risk score.

5

Borderline Value Detection

Alerts when parameters are individually normal but close to reference limits. If 3+ parameters are simultaneously borderline, early-stage pathology (e.g., early iron deficiency) is detected. In typical LIS systems, this goes completely unnoticed.

6

Explainable Output

Each output clearly shows the evaluated parameters, the relevant combination logic, the reasons for triggering, and the reference framework. It is not a black box; it is an auditable and queryable structure.

Source Types

Hematology GuidelinesWHO, BSH, ASH — anemia, iron deficiency, thrombocytopenia
Inflammation ReferencesAHA/CDC, BRINDA — CRP, ferritin-inflammation relationship
Endocrine GuidelinesEndocrine Society, ATA/AACE — Vitamin D, thyroid
Kidney GuidelinesKDIGO — chronic kidney disease anemia
CoagulationISTH — DIC scoring system
Reference TextbooksHarrison's, Williams Hematology, Wintrobe's
Peer-Reviewed JournalsNEJM, Lancet, Blood, Br J Haematol
Clinical PracticeExplainable rule sets supported by expert opinion

Temporal Clinical Trajectory Classification

HemaLens evaluates laboratory values not only as snapshots but also along the time axis. Multi-parameter trends are classified into 4 top-level clinical trajectories, and these trajectories dynamically modify the base risk score.

Deterioration2+ parameters in the same clinical group are trending toward worsening. Example: WBC↑ + CRP↑ = inflammation deterioration. Risk multiplier: x1.07 — x1.25
ImprovementParameters are trending toward normalization. Example: CRP↓ + WBC↓ = treatment response. Risk multiplier: x0.85 — x0.96
Acceleration>40% worsening rate in any parameter. Indicates an acute process. Risk multiplier: x1.05 — x1.19
SpreadingNew abnormalities in previously normal clinical groups. Suggests a systemic process. Risk multiplier: x1.04 — x1.18

Trajectory multipliers are multiplied together (not summed) to determine the final score. This allows effects to interact naturally and prevents linear stacking. Example: Deterioration (x1.175) + Acceleration (x1.1) + Spreading (x1.07) = x1.38 → If base score is 30, modified score is 41 (HIGH → CRITICAL).

Explainability Layers

HemaLens presents every decision with a multi-layered explainability framework:

Why Was It Triggered?Each parameter alert shows the triggering values, threshold comparison, and supporting contexts (e.g., how CRP↑ affects ferritin interpretation).
Why Was It Not Triggered?Expected but non-triggered patterns and their reasons are explained. Example: "B12 deficiency pattern not triggered — MCV normal, B12 normal."
Pattern ConfidenceEach combination finding is presented with a confidence percentage based on source quality and parameter concordance (Confidence: 85%, Strength: High).
Physician Feedback"Appropriate / Not Appropriate" feedback is collected on each finding block. Reasons: Too sensitive, Missing alert, Out of context. This data feeds into the rule improvement cycle.

Clinical Combinations

HemaLens simultaneously evaluates 135 explainable combinations defined in the clinical literature. Below are all categories, their associated combinations, and primary references.

Iron Metabolism & Anemia (16 combinations)
Classic Iron Deficiency Anemia
MCV↓ + RDW↑ + Ferritin↓ + HGB↓
High
Subclinical Iron Deficiency
MCV borderline + RDW↑ + Ferritin 10-29 + HGB normal
Moderate
Masked Deficiency Under Inflammation
MCV↓ + RDW↑ + Ferritin 30-50 + CRP↑
High
Functional Iron Deficiency / Chronic Disease
HGB↓ + Ferritin normal/↑ + CRP↑
High
Pre-Anemia Store Depletion
HGB normal + MCV↓ + Ferritin↓
Moderate
Mixed Anemia
MCV normal + RDW↑
Moderate
Early-Stage Anemia
HGB borderline + MCV borderline + RDW borderline + Ferritin low-normal
Moderate
Dimorphic Anemia (Dual Deficiency)
MCV normal + RDW≥18 + Ferritin↓ + B12↓ or Folate↓
High
Thalassemia Trait vs IDA (Mentzer Index)
MCV/RBC ratio + RDW
High
Low Serum Iron — Strong IDA Concordance
Serum iron↓ + Ferritin↓ + MCV↓
High
Functional Iron Deficiency (Serum Iron)
Serum iron↓ + Ferritin↑ + CRP↑
High
Low TSAT + Low Ferritin
TSAT <15% + Ferritin↓
High
Low TSAT + Inflammation
TSAT <15% + Ferritin↑ + CRP↑
High
Iron Overload Suspicion
TSAT >50% + Ferritin >300
High
Low Albumin + Anemia + Inflammation
Albumin↓ + HGB↓ + CRP↑
Moderate
GI Loss Clue (Male >50)
IDA + Male + Age >50
High

WHO. Serum ferritin concentrations. Geneva: WHO; 2020. ISBN: 9789240000124
Camaschella C. N Engl J Med. 2015;372(19):1832-1843. DOI: 10.1056/NEJMra1401038
Mentzer WC. Lancet. 1973;1(7808):882
Weiss G, et al. Blood. 2019;133(1):40-50
Lopez A, et al. Lancet. 2016;387(10021):907-916

Infection, Hematology and Sepsis (12 combinations)
Bacterial Infection Combination
WBC↑ + NEU dominant + CRP↑
High
Viral Infection / Lymphoproliferative
WBC normal/↓ + LYM↑
Moderate
Pancytopenia
HGB↓ + WBC↓ + PLT↓
High
Neutropenic Infection Risk
NEU↓ + CRP↑
High
Bacterial Sepsis
PCT >2 + WBC↑ + CRP↑
High
Sepsis Early Warning
PCT 0.5-2 + CRP↑ + NEU↑
Moderate
MDS Clue
Pancytopenia + MCV↑ + Age>60
High
Marked Lymphocytosis
LYM# >10 — lymphoproliferative suspicion
High
Moderate Lymphocytosis
LYM# 5-10 + WBC↑ — follow-up
Moderate
B12 Neurological Risk
B12 <150 + MCV >110
High
NLR + Sepsis Prognosis
NLR >5 + CRP↑ + PCT↑
High
NLR + Inflammation Follow-up
NLR >3 + CRP↑
Moderate

Surviving Sepsis Campaign 2021. DOI: 10.1007/s00134-021-06506-y
Zahorec R. Bratisl Lek Listy. 2001;102(1):5-14
Arber DA, et al. Blood. 2016;127(20):2391-2405 (WHO MDS)
Swerdlow SH, et al. WHO Classification of Lymphoid Neoplasms 2016
Devalia V, et al. Br J Haematol. 2014;166(4):496-513

Thrombocytosis, Thyroid, Macrocytosis and Erythrocytosis (13 combinations)
Reactive Thrombocytosis
PLT↑ + CRP↑
High
Clonal Thrombocytosis Suspicion
PLT↑↑ + CRP normal
High
High MPV + Thrombocytopenia
MPV >10 + PLT <100
Moderate
Hypothyroidism-Anemia Association
TSH↑ + HGB↓ or MCV↑
Moderate
Hyperthyroidism-Anemia Association
TSH↓↓ + HGB↓
Moderate
Subclinical Hypothyroidism
TSH↑ + fT4 normal
Moderate
Subclinical Hyperthyroidism
TSH↓ + fT4 normal
Moderate
Autoimmune Thyroiditis
Anti-TPO↑ + TSH abnormal
Moderate
Polycythemia Vera Suspicion
HGB↑ + HCT↑ ± WBC↑ ± PLT↑ (WHO 2016)
High
Megaloblastic Anemia
MCV >110 + B12↓ or Folate↓
High
Marked Macrocytosis
MCV >110 (B12/folate unknown)
Moderate
Hypoproliferative Anemia
Reticulocyte <1% + HGB↓
Moderate
Erythrocytosis (PV Criteria)
HGB >16.5 (M) / >16 (F) + HCT↑ (WHO 2016)
Moderate

Harrison CN, et al. N Engl J Med. 2005;353(1):33-45
Green R, Datta Mitra A. Med Clin North Am. 2017;101(1):169-193
Garber JR, et al. Thyroid. 2012;22(12):1200-1235. DOI: 10.1089/thy.2012.0205
Buber E, et al. Platelets. 2020

Kidney, Liver, GI and Coagulation (28 combinations)
Chronic Kidney Disease Anemia
GFR↓ + HGB↓
High
Hemolytic Anemia Triad
LDH↑ + Bilirubin↑ + HGB↓
High
Liver + Thrombocytopenia
ALT/AST↑ + PLT↓
High
DIC Early Warning
PLT↓ + INR↑
High
De Ritis Ratio + Anemia
AST/ALT>2 + HGB↓
Moderate
FIB-4 Advanced Fibrosis
FIB-4 >3.25 + transaminase↑
High
FIB-4 Indeterminate Fibrosis
FIB-4 1.45-3.25
Moderate
GI Loss Clue
IDA + Male + Age>50
High
BUN/Creatinine Prerenal
BUN/Crea >20
Moderate
BUN + GI Bleeding
BUN↑ + IDA
High
GGT + Transaminase
GGT↑ + ALT/AST↑
Moderate
GGT + De Ritis Alcoholic Liver
GGT↑ + AST/ALT >2
High
ALP + GGT Cholestasis
ALP↑ + GGT↑
Moderate
ALP + Osteomalacia
ALP↑ + GGT normal + VitD↓
Moderate
Hypercalcemia + Myeloma
Ca↑ + HGB↓ + ESR↑
High
Hypercalcemia + Kidney
Ca↑ + Creatinine↑
High
Hypocalcemia + Albumin
Ca↓ + Albumin↓
Moderate
TIBC + IDA
TIBC↑ + Ferritin↓
High
TIBC + Chronic Disease
TIBC↓ + Ferritin↑ + CRP↑
High
Uric Acid + TLS
Uric↑ + WBC↑ + LDH↑
High
Pre-DIC / Early DIC
PLT↓ + INR↑ + D-dimer↑ + aPTT↑
Moderate
Low Fibrinogen + DIC
Fibrinogen↓ + PLT↓ + INR↑
High
High Fibrinogen + Acute Phase
Fibrinogen↑ + CRP↑ + WBC↑
Moderate
Prolonged aPTT + Coagulopathy
aPTT↑ + PLT↓ + INR↑
High
Isolated aPTT Prolongation
aPTT↑ + INR and PLT normal
Moderate
D-dimer + DIC
D-dimer↑ + PLT↓ + INR↑
High
Decompensated Liver Disease
INR↑ + Albumin↓ + Bilirubin↑ + PLT↓
High

KDIGO. Kidney Int Suppl. 2012;2(4):279-335
Dhaliwal G, et al. Am Fam Physician. 2004;69(11):2599-2606
Taylor FB, et al. Thromb Haemost. 2001;86(5):1327-1330 (ISTH DIC)
Levi M, et al. Blood. 2018;131(8):845-854
Sterling RK, et al. Hepatology. 2006;43(6):1317-1325 (FIB-4)
Goddard AF, et al. Gut. 2011;60(10):1309-1316 (GI loss)
Botros M, Sikaris KA. Clin Biochem Rev. 2013;34(3):117-130
EASL Decompensated Cirrhosis. J Hepatol. 2018;69(2):406-460

Diabetes, Metabolism and Lipid (16 combinations)
Diabetes
HbA1c >=6.5%
High
Prediabetes
HbA1c 5.7-6.4%
Moderate
Diabetic Nephropathy
HbA1c↑ + Creatinine↑ + GFR↓
High
Diabetes + Anemia
HbA1c↑ + HGB↓
Moderate
Hypoglycemia Warning
Fasting glucose <70 mg/dL
High
Gestational Diabetes Risk
Female 18-45 + HbA1c 5.7-6.4%
Moderate
Very High TG + Diabetes
TG >500 + HbA1c↑ — pancreatitis risk
High
Metabolic Syndrome Clue
TG↑ + HbA1c↑ + Glucose↑
Moderate
Metabolic Syndrome (HOMA-IR)
TG↑ + HDL↓ + Glucose↑ + HOMA-IR >2.5
High
TG + NAFLD
TG >150 + ALT↑
Moderate
Low HDL + TG + Prediabetes
HDL↓ + TG↑ + HbA1c↑
High
High LDL + Diabetes
LDL >130 + HbA1c↑ — CV risk
High
Dyslipidemia Triad
LDL↑ + TG↑ + HDL↓
High
High Cholesterol + LDL
Total cholesterol >240 + LDL >160
Moderate
Diabetes + Dyslipidemia
Total cholesterol↑ + HbA1c↑ + TG↑
High
Low HDL + Diabetes
HDL↓ + HbA1c↑ — CV risk
Moderate

ADA. Standards of Care in Diabetes 2024. DOI: 10.2337/dc24-SINT
Alberti KG, et al. Circulation. 2009;120(16):1640-1645
Grundy SM, et al. Circulation. 2019;139(25):e1082-e1143
Berglund L, et al. J Clin Endocrinol Metab. 2012;97(9):2969-2989
Chalasani N, et al. Hepatology. 2018;67(1):328-357

Vitamin D Deficiency (2 combinations)
Severe Vitamin D Deficiency
VitD <10 ng/mL
High
Vitamin D Deficiency
VitD 10-20 ng/mL
Moderate

Holick MF, et al. J Clin Endocrinol Metab. 2011;96(7):1911-1930 (Endocrine Society Guidelines)

Pregnancy and Reproductive Health (3 combinations)
IDA in Reproductive Age
Female 18-45 + HGB <11 + MCV <80
Moderate
Low Ferritin in Reproductive Age
Female 18-45 + Ferritin <20
Moderate
Gestational Diabetes Risk
Female 18-45 + HbA1c 5.7-6.4%
Moderate

WHO Anaemia in Pregnancy Guidelines 2016
ADA Standards of Care 2024. DOI: 10.2337/dc24-SINT

Hemolysis and Autoimmune (8 combinations)
Intravascular Hemolysis
Haptoglobin↓ + LDH↑ + HGB↓
High
Hemolysis + AIHA
Haptoglobin↓ + Bilirubin↑ + HGB↓
High
Direct Coombs + Hemolysis
DAT+ + LDH↑ + HGB↓
High
Direct Coombs + Triad
DAT+ + Bilirubin↑ + HGB↓
High
Reticulocyte + Hemolysis
Retic >3% + LDH↑ + HGB↓
Moderate
Aplastic Anemia Clue
Retic <0.5% + Pancytopenia
High
Celiac + IDA
Anti-tTG+ + MCV↓ + Ferritin↓
High
Celiac Malnutrition
Anti-tTG+ + HGB↓ + Albumin↓
High

Barcellini W, et al. Transfus Med Rev. 2015;29(2):105-112
Zanella A, et al. Haematologica. 2014;99(10):1547-1554
Ludvigsson JF, et al. Gut. 2013;62(1):43-52
Killick SB, et al. Br J Haematol. 2016;172(2):187-207

Thalassemia, Genetics and MPN (4 combinations)
HbA2 + Beta Thalassemia Trait
HbA2 >3.5% + MCV↓ + RDW normal
High
Low HbA2 + Alpha Thalassemia
HbA2 <1.5% + MCV↓
Moderate
JAK2 + PV
JAK2+ + HGB↑↑
High
JAK2 + MPN
JAK2+ + PLT↑ + WBC↑
High

Galanello R, Origa R. Orphanet J Rare Dis. 2010;5:11
Arber DA, et al. Blood. 2016;127(20):2391-2405 (WHO 2016 MPN)

Hemochromatosis and Iron Overload (1 combination)
Hemochromatosis + Liver Damage
TSAT >60% + Ferritin >1000 + ALT↑
High

Bacon BR, et al. Hepatology. 2011;54(1):328-343
EASL CPG Haemochromatosis. J Hepatol. 2022;77(2):479-502

Nephrotic Syndrome (1 combination)
Nephrotic Syndrome Context
Albumin <2.5 + TP <5.0 + Creatinine↑
Moderate

KDIGO 2021 Glomerular Diseases Guideline
Kodner C. Am Fam Physician. 2009;80(10):1129-1134

Tumor Lysis Syndrome (2 combinations)
TLS Risk
Uric >10 + LDH >500 + WBC >30
High
TLS Suspicion
Uric >9 + WBC↑ + LDH↑
High

Cairo MS, Bishop M. Br J Haematol. 2004;127(1):3-11
Howard SC, et al. N Engl J Med. 2011;364(19):1844-1854

Hepatitis Serology (4 combinations)
HBsAg + Active Hepatitis B
HBsAg+ + ALT↑
High
HBsAg + Liver Disease
HBsAg+ + PLT↓ + ALT↑
High
Anti-HCV + Active Hepatitis C
Anti-HCV+ + ALT↑
High
Anti-HCV + Advanced Fibrosis
Anti-HCV+ + PLT↓ + FIB-4↑
High

EASL Clinical Practice Guidelines on HBV. J Hepatol. 2017;67(2):370-398
EASL Recommendations on HCV. J Hepatol. 2018;69(2):461-511

ESR, Protein and Myeloma (6 combinations)
Very High ESR + Anemia
ESR >100 + HGB↓ — myeloma suspicion
High
High ESR + CRP + Anemia
ESR >50 + CRP↑ + HGB↓
High
ESR + Inflammation Markers
ESR >50 + CRP >10 + WBC↑
Moderate
High Total Protein + Myeloma
TP >9.0 + ESR↑ + HGB↓
High
Low TP + Albumin
TP <6.0 + Albumin↓ — malnutrition/protein loss
Moderate
Smoldering Myeloma / MGUS
TP >8.0 + ESR >50 + HGB <13
Moderate

Kyle RA, et al. N Engl J Med. 2002;346(8):564-569
Rajkumar SV, et al. Lancet Oncol. 2014;15(12):e538-548

Blood Gas and Metabolic Emergency (1 combination)
Euglycemic DKA
HCO3↓ + Anion gap↑ + Beta-hydroxybutyrate↑
High

ADA Standards of Care 2024
Kitabchi AE, et al. Diabetes Care. 2009;32(7):1335-1343

Adrenal, Oncology and Special Patterns (5 combinations)
Primary Adrenal Insufficiency
Na↓ + K↑ + Glucose↓ + Eosinophil↑
Moderate
Chronic Inflammation / Lymphoma Suspicion
LDH↑ + ESR↑ + CRP↑ + HGB↓
Moderate
Hepatocellular Carcinoma Risk
FIB-4↑ + PLT↓ + Albumin↓
Moderate
New-Onset Diabetes + Cholestasis (Pancreas)
HbA1c↑ + ALP↑ + GGT↑ + Bilirubin↑
Moderate
TTP / TMA
PLT↓ + LDH↑ + Haptoglobin↓ + Reticulocyte↑
High

Bornstein SR, et al. J Clin Endocrinol Metab. 2016;101(2):364-389
EASL Clinical Practice Guidelines on HCC. J Hepatol. 2022;77(1):170-202
NCCN Guidelines — Pancreatic Adenocarcinoma, Version 2.2024
BSH Guidelines for TTP. Br J Haematol. 2012;158(3):323-335

Analytical Interference Detection (11 rules + 3 HIL indices)
Biotin Interference
6 tests: TSH, fT4, VitD, Ferritin, D-dimer, HbA1c
Info
Cholestasis → CA 19-9 False ↑
Bilirubin >2.0 + ALP >120
High
Hemolysis → LDH False ↑
LDH >250 + Haptoglobin <25
High
Hemolysis → Potassium False ↑
K >5.0 + Haptoglobin <25
High
Bilirubin → Creatinine False ↓
Creatinine + Bilirubin >3.0 (Jaffe method)
High
Lipemia → Total Protein False ↑
Total protein + TG >400
High
HIL H-Index (Hemolysis)
7 parameters affected: K, LDH, AST, ALT, HGB, aPTT, INR
CLSI
HIL I-Index (Icterus)
5 parameters affected: Bilirubin, Creatinine, Albumin, TP
CLSI
HIL L-Index (Lipemia)
6 parameters affected: LDH, TP, Albumin, Na, Ca, CRP
CLSI

CLSI EP7-A2. Interference Testing in Clinical Chemistry. 2005
Tramboli RP & Plebani M. Crit Rev Clin Lab Sci. 2018
Goonetilleke & Siriwardena. Eur J Surg Oncol. 2007;33(3):266-270
Stahl M, et al. Scand J Clin Lab Invest. 2000;60(4):307-316
Murthy K, et al. Clin Chem. 1998;44(3):562-568

Disease Suspicion Layer — Disease Aggregation (5 diseases)

When multiple combinations point to the same disease, the system aggregates them into a disease-level confidence score. Each disease has threshold-based criteria, negative modifiers, and collision penalties.

Multiple Myeloma
7 criteria: TP>8.0, ESR>50, HGB<13, Ca>10.5, B2M>2.4, Cr>1.2, LDH>250
IMWG / Kyle 2002
Lymphoma
6 criteria: LDH, ESR, CRP, HGB, LYM_ABS, Beta-2 MG
Harrison's
Hepatocellular Carcinoma
7 criteria: AFP, ALT, AST, PLT, Albumin, Bilirubin, ALP
EASL HCC 2022
Pancreatic Pathology
7 criteria: HbA1c>6.5, ALP>120, GGT>80, Bil>1.5, CA19-9>37, Amylase>125, Lipase>140
NCCN 2024
Metabolic Syndrome
6 criteria: TG, HDL, Glucose, HOMA-IR, HbA1c, ALT
IDF 2006 / ATP III

Confidence = lookup_table[active_criteria] × (1 - negative_penalty)
Negative modifier: criteria not meeting threshold automatically added as risk-reducing factor
Collision penalty: low-specificity (<0.5) shared patterns reduce by 5%/pattern

Parameter Threshold Values and Sources
Hemoglobin
Female <12, Male <13 g/dL = anemia
WHO 2011/2024
Ferritin
Healthy adult <15 µg/L = deficiency. With inflammation <70
WHO 2020
Vitamin B12
<148 pmol/L (≈200 pg/mL) = deficiency likely
BSH 2014
CRP
<3 low risk, >5 inflammation (BRINDA), >10 acute phase
AHA/CDC 2003
TSH
0.4-4.12 normal, >10 treatment recommended
ATA/AACE 2012
GFR
<60 = chronic kidney disease stage 3+
KDIGO 2012
Mentzer Index
MCV/RBC: <13 thalassemia, >13 IDA
Mentzer 1973
BUN
7-20 mg/dL, BUN/Crea >20 = prerenal
Srygley 2012
Uric Acid
M: 3.4-7.0, F: 2.4-6.0 mg/dL
Richette 2010
GGT
<55 U/L, elevation indicates alcohol or cholestasis
Botros 2013
ALP
<120 U/L, elevation indicates liver or bone
EASL 2018
Calcium
8.5-10.5 mg/dL
Bilezikian 2014
TIBC
250-400 µg/dL, >400 = IDA
WHO 2017
JAK2 V617F
Positive = MPN suspicion
WHO 2016
HbA2
1.5-3.5%, >3.5 = beta thalassemia trait
Galanello 2010
Anti-tTG
<10 U/mL negative, >10 = celiac suspicion
Ludvigsson 2013
Transparency Commitment HemaLens is an explainable system. Each output clearly shows the evaluated parameters, the relevant combination logic, the reasons for triggering and non-triggering, the temporal trajectory classification, and the reference framework used. The system is not a black box; every rule is auditable, queryable, and improvable through physician feedback.
This software is used within the scope of a pilot program and clinical research. It does not diagnose or recommend treatment. The final clinical decision belongs entirely to the physician.