{
 "meta": {
  "figures": 400,
  "seed_range": [
   7000,
   7399
  ],
  "engine_version": "0.1.0",
  "generated_at": "2026-09-25T00:29:45+00:00",
  "failures": {
   "DigitizeError:  x-axis tick labels are not consistent with a linear axis.": 1,
   "DigitizeError:  No survival curves could be traced.": 1
  },
  "source": "evals/results/final_heldout.json"
 },
 "summary": {
  "figures": 400,
  "success_rate": 0.995,
  "arm_count_correct": 0.9396984924623115,
  "curve_iae": {
   "n": 824,
   "mean": 0.014531577552433355,
   "median": 0.002611152938091575,
   "p90": 0.005444379414254252,
   "max": 0.8268109951557935
  },
  "curve_max_err": {
   "n": 824,
   "mean": 0.05585422606614262,
   "median": 0.027216683257865193,
   "p90": 0.08249722402919772,
   "max": 1.0
  },
  "median_abs_err_months": {
   "n": 638,
   "mean": 0.6138380877742946,
   "median": 0.08289999999999953,
   "p90": 0.6592299999999981,
   "max": 41.0565
  },
  "median_rel_err": {
   "n": 638,
   "mean": 0.02925973112232353,
   "median": 0.004462558049535658,
   "p90": 0.034259628073422324,
   "max": 3.830394524959742
  },
  "events_rel_err": {
   "n": 786,
   "mean": 0.09289901966035799,
   "median": 0.011235955056179775,
   "p90": 0.2610753502461189,
   "max": 3.282051282051282
  },
  "rmst_rel_err": {
   "n": 786,
   "mean": 0.008692397856852434,
   "median": 0.0010702062599236301,
   "p90": 0.0040503524027389335,
   "max": 0.7027539459142035
  },
  "abs_log_hr_err": {
   "n": 357,
   "mean": 0.08388514837776984,
   "median": 0.011301710849710211,
   "p90": 0.09192775810460398,
   "max": 4.275625040281037
  },
  "hr_within_5pct": 0.8161559888579387,
  "hr_within_10pct": 0.8969359331476323,
  "true_hr_in_reconstructed_ci": 0.9693593314763231,
  "nar_cell_accuracy": 0.9248473461719117,
  "nar_cell_error_rate": 0.028338813214341632,
  "arm_name_accuracy": 0.9004854368932039,
  "secs_per_figure": {
   "n": 398,
   "mean": 11.484401719054985,
   "median": 11.51937186717987,
   "p90": 15.891188240051271,
   "max": 22.70752501487732
  },
  "failures": {
   "DigitizeError:  x-axis tick labels are not consistent with a linear axis.": 1,
   "DigitizeError:  No survival curves could be traced.": 1
  }
 },
 "groups": {
  "by_palette_type": {
   "black&white": {
    "figures": 31,
    "success_rate": 0.967741935483871,
    "arm_count_correct": 0.9,
    "curve_iae": {
     "n": 63,
     "mean": 0.08184997340446053,
     "median": 0.005040107843137232,
     "p90": 0.426836484746158,
     "max": 0.6673266755888602
    },
    "curve_max_err": {
     "n": 63,
     "mean": 0.18823207516350865,
     "median": 0.04486052648258851,
     "p90": 0.7913759380373091,
     "max": 0.9870901487683237
    },
    "median_abs_err_months": {
     "n": 53,
     "mean": 3.50583962264151,
     "median": 0.15349999999999753,
     "p90": 13.24010000000002,
     "max": 38.0588
    },
    "median_rel_err": {
     "n": 53,
     "mean": 0.19036756588148881,
     "median": 0.010581092801387675,
     "p90": 0.7826509313946304,
     "max": 3.830394524959742
    },
    "events_rel_err": {
     "n": 59,
     "mean": 0.3829919927423718,
     "median": 0.05263157894736842,
     "p90": 1.0889903329752955,
     "max": 3.282051282051282
    },
    "rmst_rel_err": {
     "n": 59,
     "mean": 0.07665756036850979,
     "median": 0.0026382306645127995,
     "p90": 0.3978466856621639,
     "max": 0.6682015858688056
    },
    "abs_log_hr_err": {
     "n": 23,
     "mean": 0.692865216918574,
     "median": 0.04796261936674562,
     "p90": 3.2254727203436557,
     "max": 3.805799921160375
    },
    "hr_within_5pct": 0.48,
    "hr_within_10pct": 0.52,
    "true_hr_in_reconstructed_ci": 0.68,
    "nar_cell_accuracy": 0.7345537757437071,
    "nar_cell_error_rate": 0.17391304347826086,
    "arm_name_accuracy": 0.7777777777777778,
    "secs_per_figure": {
     "n": 30,
     "mean": 10.941993387540181,
     "median": 9.9730966091156,
     "p90": 16.30511724948883,
     "max": 21.72493600845337
    },
    "failures": {
     "DigitizeError:  No survival curves could be traced.": 1
    }
   },
   "colour": {
    "figures": 369,
    "success_rate": 0.997289972899729,
    "arm_count_correct": 0.9429347826086957,
    "curve_iae": {
     "n": 761,
     "mean": 0.008958569748651864,
     "median": 0.0025507667728237966,
     "p90": 0.004799920393297178,
     "max": 0.8268109951557935
    },
    "curve_max_err": {
     "n": 761,
     "mean": 0.04489521884783241,
     "median": 0.026554858764928246,
     "p90": 0.0737815278787034,
     "max": 1.0
    },
    "median_abs_err_months": {
     "n": 585,
     "mean": 0.35182769230769223,
     "median": 0.08170000000000144,
     "p90": 0.5450600000000003,
     "max": 41.0565
    },
    "median_rel_err": {
     "n": 585,
     "mean": 0.014663636691151283,
     "median": 0.004202258785759313,
     "p90": 0.02939595332529832,
     "max": 0.980641077698426
    },
    "events_rel_err": {
     "n": 727,
     "mean": 0.06935639873623307,
     "median": 0.010309278350515464,
     "p90": 0.2110874200426438,
     "max": 2.6
    },
    "rmst_rel_err": {
     "n": 727,
     "mean": 0.003176655644764698,
     "median": 0.0010307245585943598,
     "p90": 0.0034810993931476933,
     "max": 0.7027539459142035
    },
    "abs_log_hr_err": {
     "n": 334,
     "mean": 0.04194939515490009,
     "median": 0.010960218285190594,
     "p90": 0.06951463213139732,
     "max": 4.275625040281037
    },
    "hr_within_5pct": 0.8413173652694611,
    "hr_within_10pct": 0.9251497005988024,
    "true_hr_in_reconstructed_ci": 0.9910179640718563,
    "nar_cell_accuracy": 0.9388235294117647,
    "nar_cell_error_rate": 0.01764705882352941,
    "arm_name_accuracy": 0.9106438896189225,
    "secs_per_figure": {
     "n": 368,
     "mean": 11.528619789558908,
     "median": 11.560037016868591,
     "p90": 15.778330540657045,
     "max": 22.70752501487732
    },
    "failures": {
     "DigitizeError:  x-axis tick labels are not consistent with a linear axis.": 1
    }
   }
  },
  "by_image_quality": {
   "clean": {
    "figures": 266,
    "success_rate": 1.0,
    "arm_count_correct": 0.9473684210526315,
    "curve_iae": {
     "n": 550,
     "mean": 0.008729275602966845,
     "median": 0.0024192950212566693,
     "p90": 0.004416319934511782,
     "max": 0.7745452606060607
    },
    "curve_max_err": {
     "n": 550,
     "mean": 0.04299680390058682,
     "median": 0.02577043020463376,
     "p90": 0.06887505913333243,
     "max": 1.0
    },
    "median_abs_err_months": {
     "n": 421,
     "mean": 0.21141781472684082,
     "median": 0.08099999999999952,
     "p90": 0.45830000000000126,
     "max": 7.534800000000001
    },
    "median_rel_err": {
     "n": 421,
     "mean": 0.009677888561195029,
     "median": 0.0039165253989595005,
     "p90": 0.020842299415923787,
     "max": 0.2630314878167982
    },
    "events_rel_err": {
     "n": 526,
     "mean": 0.05925064538518176,
     "median": 0.009260053159564435,
     "p90": 0.2,
     "max": 0.9130434782608695
    },
    "rmst_rel_err": {
     "n": 526,
     "mean": 0.0015798878783228258,
     "median": 0.0009936323188770096,
     "p90": 0.0029164112207851387,
     "max": 0.029525377542080036
    },
    "abs_log_hr_err": {
     "n": 244,
     "mean": 0.023582590670967837,
     "median": 0.009324907478407835,
     "p90": 0.05819074847947954,
     "max": 0.19933006627267869
    },
    "hr_within_5pct": 0.860655737704918,
    "hr_within_10pct": 0.9426229508196722,
    "true_hr_in_reconstructed_ci": 1.0,
    "nar_cell_accuracy": 0.9648956356736242,
    "nar_cell_error_rate": 0.0009487666034155598,
    "arm_name_accuracy": 0.9218181818181819,
    "secs_per_figure": {
     "n": 266,
     "mean": 11.625655437770643,
     "median": 11.843843460083008,
     "p90": 15.930172443389893,
     "max": 22.70752501487732
    },
    "failures": {}
   },
   "degraded (downscaled/JPEG)": {
    "figures": 134,
    "success_rate": 0.9850746268656716,
    "arm_count_correct": 0.9242424242424242,
    "curve_iae": {
     "n": 274,
     "mean": 0.026178534020340576,
     "median": 0.0032227757538510755,
     "p90": 0.01810915811411228,
     "max": 0.8268109951557935
    },
    "curve_max_err": {
     "n": 274,
     "mean": 0.08166292019408308,
     "median": 0.03084002142917782,
     "p90": 0.12700727354818517,
     "max": 0.9870901487683237
    },
    "median_abs_err_months": {
     "n": 217,
     "mean": 1.394570506912442,
     "median": 0.08990000000000009,
     "p90": 2.6350600000000015,
     "max": 41.0565
    },
    "median_rel_err": {
     "n": 217,
     "mean": 0.06725031046902905,
     "median": 0.005578106586355808,
     "p90": 0.09075531255474795,
     "max": 3.830394524959742
    },
    "events_rel_err": {
     "n": 260,
     "mean": 0.1609722691555222,
     "median": 0.016597796143250687,
     "p90": 0.5469339622641506,
     "max": 3.282051282051282
    },
    "rmst_rel_err": {
     "n": 260,
     "mean": 0.02308155265957003,
     "median": 0.0013753159269692251,
     "p90": 0.01409438147807397,
     "max": 0.7027539459142035
    },
    "abs_log_hr_err": {
     "n": 113,
     "mean": 0.2140959809482095,
     "median": 0.016674949619944424,
     "p90": 0.21228975973397943,
     "max": 4.275625040281037
    },
    "hr_within_5pct": 0.7217391304347827,
    "hr_within_10pct": 0.8,
    "true_hr_in_reconstructed_ci": 0.9043478260869565,
    "nar_cell_accuracy": 0.8470750806080147,
    "nar_cell_error_rate": 0.08152924919391985,
    "arm_name_accuracy": 0.8576642335766423,
    "secs_per_figure": {
     "n": 132,
     "mean": 11.199754073764339,
     "median": 10.883824467658997,
     "p90": 15.761126565933228,
     "max": 21.72493600845337
    },
    "failures": {
     "DigitizeError:  x-axis tick labels are not consistent with a linear axis.": 1,
     "DigitizeError:  No survival curves could be traced.": 1
    }
   }
  },
  "by_risk_table": {
   "no table (N supplied)": {
    "figures": 53,
    "success_rate": 0.9811320754716981,
    "arm_count_correct": 0.8269230769230769,
    "curve_iae": {
     "n": 103,
     "mean": 0.009779428089558267,
     "median": 0.002119397889126748,
     "p90": 0.0038467655328621683,
     "max": 0.6673266755888602
    },
    "curve_max_err": {
     "n": 103,
     "mean": 0.04784191329700799,
     "median": 0.02618352195137086,
     "p90": 0.06498767024350136,
     "max": 0.9870901487683237
    },
    "median_abs_err_months": {
     "n": 84,
     "mean": 0.7129440476190477,
     "median": 0.07710000000000061,
     "p90": 0.4947399999999996,
     "max": 38.0588
    },
    "median_rel_err": {
     "n": 84,
     "mean": 0.05776476446427791,
     "median": 0.003912865473754645,
     "p90": 0.022960943854382763,
     "max": 3.830394524959742
    },
    "events_rel_err": {
     "n": 103,
     "mean": 0.3171714861903718,
     "median": 0.24311926605504589,
     "p90": 0.6783724810678986,
     "max": 1.7123287671232876
    },
    "rmst_rel_err": {
     "n": 103,
     "mean": 0.010028991695074218,
     "median": 0.0007796547827388119,
     "p90": 0.002112324880538341,
     "max": 0.6682015858688056
    },
    "abs_log_hr_err": {
     "n": 48,
     "mean": 0.09870831141107078,
     "median": 0.06575898226476386,
     "p90": 0.17536679613302406,
     "max": 0.8991722911523862
    },
    "hr_within_5pct": 0.3877551020408163,
    "hr_within_10pct": 0.6326530612244898,
    "true_hr_in_reconstructed_ci": 0.9387755102040817,
    "nar_cell_accuracy": null,
    "nar_cell_error_rate": null,
    "arm_name_accuracy": 0.7087378640776699,
    "secs_per_figure": {
     "n": 52,
     "mean": 6.024748279498174,
     "median": 5.893233895301819,
     "p90": 7.339950609207153,
     "max": 9.162492990493774
    },
    "failures": {
     "DigitizeError:  x-axis tick labels are not consistent with a linear axis.": 1
    }
   },
   "with at-risk table": {
    "figures": 347,
    "success_rate": 0.9971181556195965,
    "arm_count_correct": 0.9566473988439307,
    "curve_iae": {
     "n": 721,
     "mean": 0.015210456047129792,
     "median": 0.0026977806515335297,
     "p90": 0.005650748291896313,
     "max": 0.8268109951557935
    },
    "curve_max_err": {
     "n": 721,
     "mean": 0.056998842176018985,
     "median": 0.027467483586250563,
     "p90": 0.08693936062860227,
     "max": 1.0
    },
    "median_abs_err_months": {
     "n": 554,
     "mean": 0.5988111913357401,
     "median": 0.08329999999999949,
     "p90": 0.675960000000001,
     "max": 41.0565
    },
    "median_rel_err": {
     "n": 554,
     "mean": 0.024937668305131885,
     "median": 0.004552237518460277,
     "p90": 0.03507409960761065,
     "max": 0.980641077698426
    },
    "events_rel_err": {
     "n": 683,
     "mean": 0.05907754959799866,
     "median": 0.008403361344537815,
     "p90": 0.04691876750700304,
     "max": 3.282051282051282
    },
    "rmst_rel_err": {
     "n": 683,
     "mean": 0.008490832461044463,
     "median": 0.0011213008841797361,
     "p90": 0.004617887577009156,
     "max": 0.7027539459142035
    },
    "abs_log_hr_err": {
     "n": 309,
     "mean": 0.08158252111046094,
     "median": 0.009513456142473564,
     "p90": 0.050602566838793904,
     "max": 4.275625040281037
    },
    "hr_within_5pct": 0.8838709677419355,
    "hr_within_10pct": 0.9387096774193548,
    "true_hr_in_reconstructed_ci": 0.9741935483870968,
    "nar_cell_accuracy": 0.9248473461719117,
    "nar_cell_error_rate": 0.028338813214341632,
    "arm_name_accuracy": 0.9278779472954231,
    "secs_per_figure": {
     "n": 346,
     "mean": 12.30492766950861,
     "median": 12.040466547012329,
     "p90": 16.251598000526428,
     "max": 22.70752501487732
    },
    "failures": {
     "DigitizeError:  No survival curves could be traced.": 1
    }
   }
  },
  "by_ci_bands": {
   "CI bands": {
    "figures": 59,
    "success_rate": 1.0,
    "arm_count_correct": 0.9152542372881356,
    "curve_iae": {
     "n": 125,
     "mean": 0.032991908888967476,
     "median": 0.003970496408312998,
     "p90": 0.05695653498149365,
     "max": 0.6449605068892411
    },
    "curve_max_err": {
     "n": 125,
     "mean": 0.10345266312459572,
     "median": 0.03227889548025703,
     "p90": 0.265847265893782,
     "max": 0.9802662962962961
    },
    "median_abs_err_months": {
     "n": 97,
     "mean": 1.4930577319587623,
     "median": 0.12769999999999992,
     "p90": 3.8239799999999984,
     "max": 25.448
    },
    "median_rel_err": {
     "n": 97,
     "mean": 0.06648959574185209,
     "median": 0.007586206896551694,
     "p90": 0.1241578833483715,
     "max": 0.9479962471518564
    },
    "events_rel_err": {
     "n": 116,
     "mean": 0.22283722695441574,
     "median": 0.018018180506456965,
     "p90": 0.712474645030426,
     "max": 3.282051282051282
    },
    "rmst_rel_err": {
     "n": 116,
     "mean": 0.0284167749289527,
     "median": 0.0017025279431156528,
     "p90": 0.038768503723621464,
     "max": 0.6366397354332753
    },
    "abs_log_hr_err": {
     "n": 48,
     "mean": 0.2555953550279268,
     "median": 0.01755258802663049,
     "p90": 0.47644880739760564,
     "max": 3.623888106089328
    },
    "hr_within_5pct": 0.673469387755102,
    "hr_within_10pct": 0.6938775510204082,
    "true_hr_in_reconstructed_ci": 0.8775510204081632,
    "nar_cell_accuracy": 0.8134920634920635,
    "nar_cell_error_rate": 0.11607142857142858,
    "arm_name_accuracy": 0.816,
    "secs_per_figure": {
     "n": 59,
     "mean": 11.308097204919589,
     "median": 10.610107183456421,
     "p90": 16.412840843200687,
     "max": 21.72493600845337
    },
    "failures": {}
   },
   "no CI bands": {
    "figures": 341,
    "success_rate": 0.9941348973607038,
    "arm_count_correct": 0.943952802359882,
    "curve_iae": {
     "n": 699,
     "mean": 0.01123037380841795,
     "median": 0.002491205751335949,
     "p90": 0.004616897937198636,
     "max": 0.8268109951557935
    },
    "curve_max_err": {
     "n": 699,
     "mean": 0.047342345333228966,
     "median": 0.026363441690065895,
     "p90": 0.07362070651569995,
     "max": 1.0
    },
    "median_abs_err_months": {
     "n": 541,
     "mean": 0.45619611829944545,
     "median": 0.08170000000000144,
     "p90": 0.5122999999999998,
     "max": 41.0565
    },
    "median_rel_err": {
     "n": 541,
     "mean": 0.022584505857823945,
     "median": 0.00422072379651745,
     "p90": 0.02303200000000004,
     "max": 3.830394524959742
    },
    "events_rel_err": {
     "n": 670,
     "mean": 0.07040225541243157,
     "median": 0.01020408163265306,
     "p90": 0.21442577030812326,
     "max": 3.0
    },
    "rmst_rel_err": {
     "n": 670,
     "mean": 0.0052774310801903005,
     "median": 0.0010162278006482176,
     "p90": 0.0033783381827450394,
     "max": 0.7027539459142035
    },
    "abs_log_hr_err": {
     "n": 309,
     "mean": 0.05721171821852214,
     "median": 0.010408921245119662,
     "p90": 0.06581370630029797,
     "max": 4.275625040281037
    },
    "hr_within_5pct": 0.8387096774193549,
    "hr_within_10pct": 0.9290322580645162,
    "true_hr_in_reconstructed_ci": 0.9838709677419355,
    "nar_cell_accuracy": 0.9457148168804611,
    "nar_cell_error_rate": 0.01189812232757018,
    "arm_name_accuracy": 0.9155937052932761,
    "secs_per_figure": {
     "n": 339,
     "mean": 11.515085985526927,
     "median": 11.569576978683472,
     "p90": 15.77285804748535,
     "max": 22.70752501487732
    },
    "failures": {
     "DigitizeError:  x-axis tick labels are not consistent with a linear axis.": 1,
     "DigitizeError:  No survival curves could be traced.": 1
    }
   }
  },
  "by_arms": {
   "2 arms": {
    "figures": 354,
    "success_rate": 0.9943502824858758,
    "arm_count_correct": 0.9375,
    "curve_iae": {
     "n": 688,
     "mean": 0.013631038351213263,
     "median": 0.002544522991598351,
     "p90": 0.004824386962503849,
     "max": 0.7745452606060607
    },
    "curve_max_err": {
     "n": 688,
     "mean": 0.05378525051864622,
     "median": 0.026900572571892087,
     "p90": 0.07931649713525603,
     "max": 1.0
    },
    "median_abs_err_months": {
     "n": 529,
     "mean": 0.441781663516068,
     "median": 0.08099999999999952,
     "p90": 0.5366999999999983,
     "max": 38.0588
    },
    "median_rel_err": {
     "n": 529,
     "mean": 0.025875569220244317,
     "median": 0.004286478859533924,
     "p90": 0.029314032719345184,
     "max": 3.830394524959742
    },
    "events_rel_err": {
     "n": 650,
     "mean": 0.08615374797287531,
     "median": 0.011131763733994217,
     "p90": 0.2508992805755398,
     "max": 2.6
    },
    "rmst_rel_err": {
     "n": 650,
     "mean": 0.007153510979279972,
     "median": 0.0010014639530516378,
     "p90": 0.0034541830994357084,
     "max": 0.6682015858688056
    },
    "abs_log_hr_err": {
     "n": 312,
     "mean": 0.07253638547254702,
     "median": 0.011535379464186798,
     "p90": 0.08420805537616083,
     "max": 3.805799921160375
    },
    "hr_within_5pct": 0.8184713375796179,
    "hr_within_10pct": 0.9044585987261147,
    "true_hr_in_reconstructed_ci": 0.9713375796178344,
    "nar_cell_accuracy": 0.926416906820365,
    "nar_cell_error_rate": 0.017483189241114312,
    "arm_name_accuracy": 0.8997093023255814,
    "secs_per_figure": {
     "n": 352,
     "mean": 11.068625498224389,
     "median": 11.109102964401245,
     "p90": 15.303738403320322,
     "max": 19.77340269088745
    },
    "failures": {
     "DigitizeError:  x-axis tick labels are not consistent with a linear axis.": 1,
     "DigitizeError:  No survival curves could be traced.": 1
    }
   },
   "3 arms": {
    "figures": 46,
    "success_rate": 1.0,
    "arm_count_correct": 0.9565217391304348,
    "curve_iae": {
     "n": 136,
     "mean": 0.019087246452723222,
     "median": 0.0032699246852559505,
     "p90": 0.011525021897693971,
     "max": 0.8268109951557935
    },
    "curve_max_err": {
     "n": 136,
     "mean": 0.06632080824759498,
     "median": 0.028930039876798957,
     "p90": 0.10734858912315637,
     "max": 0.9802453448678979
    },
    "median_abs_err_months": {
     "n": 109,
     "mean": 1.4488642201834865,
     "median": 0.1095000000000006,
     "p90": 2.8872800000000005,
     "max": 41.0565
    },
    "median_rel_err": {
     "n": 109,
     "mean": 0.04568378292232263,
     "median": 0.004982294325949197,
     "p90": 0.09439559107771209,
     "max": 0.980641077698426
    },
    "events_rel_err": {
     "n": 136,
     "mean": 0.12513745051965017,
     "median": 0.011561079925153702,
     "p90": 0.2847824553980678,
     "max": 3.282051282051282
    },
    "rmst_rel_err": {
     "n": 136,
     "mean": 0.01604737190407376,
     "median": 0.0015204061681988869,
     "p90": 0.008104354950902459,
     "max": 0.7027539459142035
    },
    "abs_log_hr_err": {
     "n": 45,
     "mean": 0.1625699045206479,
     "median": 0.009575194954615007,
     "p90": 0.13445631813598907,
     "max": 4.275625040281037
    },
    "hr_within_5pct": 0.8,
    "hr_within_10pct": 0.8444444444444444,
    "true_hr_in_reconstructed_ci": 0.9555555555555556,
    "nar_cell_accuracy": 0.9179357021996616,
    "nar_cell_error_rate": 0.07614213197969544,
    "arm_name_accuracy": 0.9044117647058824,
    "secs_per_figure": {
     "n": 46,
     "mean": 14.665993669758672,
     "median": 14.734307885169983,
     "p90": 20.45500099658966,
     "max": 22.70752501487732
    },
    "failures": {}
   }
  }
 }
}