19  Core Level Analysis

20 Core Level Analysis

This chapter provides detailed core-level analysis, examining diagnostic agreement, duration, and performance metrics at the individual biopsy core level rather than aggregated by case.

20.1 Core Level Diagnosis Distribution

Note for Pathologists: This table shows the distribution of diagnoses (Benign, Malignant, ASAP/Consult) at the individual biopsy core level for each pathologist, comparing ‘Without AI’ and ‘With AI’.

Core-Level Diagnosis Distribution (%)
Pathologist Count With AI_Benign With AI_Malignant With AI_Consult With AI_IHC Without AI_Benign Without AI_Malignant Without AI_Consult Without AI_IHC
P1 567 70.1 0.0 0.0 0.0 0.0 0.0 0.0 0.0
P1 197 0.0 24.4 0.0 0.0 0.0 0.0 0.0 0.0
P1 11 0.0 0.0 1.4 0.0 0.0 0.0 0.0 0.0
P1 34 0.0 0.0 0.0 4.2 0.0 0.0 0.0 0.0
P1 488 0.0 0.0 0.0 0.0 60.3 0.0 0.0 0.0
P1 184 0.0 0.0 0.0 0.0 0.0 22.7 0.0 0.0
P1 22 0.0 0.0 0.0 0.0 0.0 0.0 2.7 0.0
P1 115 0.0 0.0 0.0 0.0 0.0 0.0 0.0 14.2
P2 590 72.8 0.0 0.0 0.0 0.0 0.0 0.0 0.0
P2 202 0.0 24.9 0.0 0.0 0.0 0.0 0.0 0.0
P2 18 0.0 0.0 0.0 2.2 0.0 0.0 0.0 0.0
P2 579 0.0 0.0 0.0 0.0 71.6 0.0 0.0 0.0
P2 194 0.0 0.0 0.0 0.0 0.0 24.0 0.0 0.0
P2 36 0.0 0.0 0.0 0.0 0.0 0.0 0.0 4.4
P3 596 73.6 0.0 0.0 0.0 0.0 0.0 0.0 0.0
P3 194 0.0 24.0 0.0 0.0 0.0 0.0 0.0 0.0
P3 20 0.0 0.0 0.0 2.5 0.0 0.0 0.0 0.0
P3 592 0.0 0.0 0.0 0.0 73.1 0.0 0.0 0.0
P3 176 0.0 0.0 0.0 0.0 0.0 21.7 0.0 0.0
P3 6 0.0 0.0 0.0 0.0 0.0 0.0 0.7 0.0
P3 36 0.0 0.0 0.0 0.0 0.0 0.0 0.0 4.4
P4 591 73.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
P4 200 0.0 24.7 0.0 0.0 0.0 24.7 0.0 0.0
P4 19 0.0 0.0 0.0 2.3 0.0 0.0 0.0 0.0
P4 528 0.0 0.0 0.0 0.0 65.3 0.0 0.0 0.0
P4 1 0.0 0.0 0.0 0.0 0.0 0.0 0.1 0.0
P4 80 0.0 0.0 0.0 0.0 0.0 0.0 0.0 9.9

20.2 Core Level Agreement Analysis

20.2.1 Inter-Pathologist Agreement (Without AI)

Core-Level Agreement Among Pathologists

Analyzing diagnostic agreement at the core level without AI assistance.

Note for Pathologists: Agreement statistics (Kappa, % Agreement) for diagnoses at the core level between all four pathologists (Unassisted).

### Agreement Statistics (Without AI)

 INTERRATER RELIABILITY

 Interrater Reliability                        
 ───────────────────────────────────────────── 
                  Fleiss' Kappa for m Raters   
 ───────────────────────────────────────────── 
   Subjects                              807   
   Raters                                  4   
   Agreement %                      73.97770   
   Kappa                           0.6897420   
   z                                61.38201   
   p-value        < .0000001   
 ─────────────────────────────────────────────

20.2.2 Inter-Pathologist Agreement (With AI)

Impact of AI on Core-Level Agreement

Comparing inter-pathologist agreement when AI assistance is available.

Note for Pathologists: Agreement statistics for diagnoses at the core level with AI assistance.

### Agreement Statistics (With AI)

 INTERRATER RELIABILITY

 Interrater Reliability                        
 ───────────────────────────────────────────── 
                  Fleiss' Kappa for m Raters   
 ───────────────────────────────────────────── 
   Subjects                              809   
   Raters                                  4   
   Agreement %                      91.71817   
   Kappa                           0.8850523   
   z                                68.81724   
   p-value        < .0000001   
 ─────────────────────────────────────────────

20.3 Core Level Duration Analysis

20.3.1 Distribution of Diagnosis Times

Note for Pathologists: Summary of time taken (in minutes) to diagnose each core.

Core-Level Diagnosis Duration Summary (Minutes)
Pathologist AI_Status N Mean SD Median Q25 Q75
P1 With AI 781 0.76 0.51 0.59 0.38 1.00
P1 Without AI 781 0.65 0.48 0.51 0.34 0.81
P2 With AI 781 0.73 0.47 0.59 0.42 0.90
P2 Without AI 781 0.78 0.51 0.67 0.47 0.95
P3 With AI 781 0.67 0.50 0.48 0.35 0.82
P3 Without AI 781 0.77 0.46 0.61 0.42 1.01
P4 With AI 781 0.83 0.61 0.59 0.44 1.02
P4 Without AI 781 0.76 0.59 0.56 0.38 0.93

20.4 Core Level Gleason Scoring

20.4.1 Primary and Secondary Pattern Distribution

Note for Pathologists: Distribution of Primary Gleason Patterns assigned to malignant cores.

### Primary Pattern Distribution
Primary Gleason Pattern Distribution (Core Level)
Pathologist With AI_3 With AI_4 With AI_5 Without AI_3 Without AI_4 Without AI_5
P1 85 62 20 91 40 36
P2 72 89 6 63 86 18
P3 76 83 8 46 77 44
P4 66 93 8 37 118 12

Note for Pathologists: Distribution of Secondary Gleason Patterns assigned to malignant cores.


### Secondary Pattern Distribution
Secondary Gleason Pattern Distribution (Core Level)
Pathologist With AI_3 With AI_4 With AI_5 Without AI_3 Without AI_4 Without AI_5
P1 54 79 34 84 48 35
P2 39 113 15 67 85 15
P3 37 90 40 40 68 59
P4 59 91 17 44 104 19

20.5 Core Level Tumor Percentage

Note for Pathologists: Summary of tumor percentage estimates for malignant cores.

Tumor Percentage Summary at Core Level
Pathologist AI_Status N Mean SD Median Q25 Q75
P1 With AI 19 45150.4 82.7 45219 45056 45219
P1 Without AI 25 45114.7 79.9 45056 45056 45219
P2 With AI 22 45189.4 64.3 45219 45219 45219
P2 Without AI 18 45092.2 69.7 45056 45056 45056
P3 With AI 32 45122.2 81.3 45056 45056 45219
P3 Without AI 37 45130.9 82.4 45056 45056 45219
P4 With AI 31 45145.4 82.5 45219 45056 45219
P4 Without AI 28 45149.1 82.1 45219 45056 45219

20.6 Core Level PNI Analysis

Note for Pathologists: Rate of Perineural Invasion (PNI) detection at the core level.

PNI Detection Rate at Core Level
Pathologist AI_Status Total_Cores PNI_Positive PNI_Rate
P1 With AI 197 40 20.3
P1 Without AI 184 42 22.8
P2 With AI 201 30 14.9
P2 Without AI 195 22 11.3
P3 With AI 772 36 4.7
P3 Without AI 807 28 3.5
P4 With AI 201 42 20.9
P4 Without AI 202 30 14.9

20.7 Summary

Core-Level Analysis Key Findings:

  1. Diagnosis Distribution: Detailed breakdown of diagnoses at core level
  2. Agreement: Inter-pathologist agreement with and without AI
  3. Duration: Time spent per core with AI impact
  4. Gleason Patterns: Distribution of primary and secondary patterns
  5. Tumor Percentage: Quantitative assessment of tumor burden
  6. PNI Detection: Perineural invasion rates at core level

This core-level granularity provides insights into: - Individual core complexity and diagnostic challenges - Pathologist consistency across multiple cores - AI impact on specific diagnostic features - Quality and completeness of biopsy sampling