Machine Learning-Based Prognostic Prediction Models of Non-Metastatic Colon Cancer: Analyses Based on Surveillance, Epidemiology and End Results Database and a Chinese Cohort

Figures & data

Figure 1 Flow chart of study design.

Table 1 Study Outcomes of the CSS Model

		1-Year Cohort (n=15,164)	3-Year Cohort (n=14,967)	5-Year Cohort (n=14,356)
SEER dataset	Death Status
	Alive	14,791 (97.54%)	13,414 (89.62%)	11,977 (80.08%)
	Dead	373 (2.46%)	1553 (10.38%)	2379 (19.92%)
	CSS, months
	Mean (SD)	72.2 (24.3)	72.8 (23.8)	73.6 (24.0)
	Median [min, max]	76 [4, 107]	76 [4, 107]	77 [4, 107]
		1-Year Cohort (n=300)	3-Year Cohort (n=197)	5-Year Cohort (n=136)
External validation	Death Status
	Alive	295 (98.3%)	161 (81.73%)	86 (63.23%)
	Dead	5 (1.7%)	36 (18.27%)	50 (36.76%)
	CSS, months
	Mean (SD)	45.7 (24.3)	55.7 (24.1)	59.5 (27.8)
	Median [min, max]	39.0 [6, 186]	54.0[6, 186]	68.5[6, 186]

Table 2 Study Outcome of the R&M Model

y	1-Year Cohort (n=308)	3-Year Cohort (n=236)	5-Year Cohort (n=207)
Status (recurrence and/or metastasis)
Yes	80 (25.97%)	142 (60.17%)	149 (71.98%)
No	228 (74.03%)	94 (39.83%)	58 (28.02%)
DFS, months
Mean (SD)	31.8 (25.2)	33.6 (28.4)	31.8 (29.8)
Median [min, max]	24 [1, 110]	22 [1, 110]	18 [1, 110]

Figure 2 (A) Flow chart of included patients (SEER database). (B) Flow chart of included patients (Xiyuan Hospital in Beijing, China).

Figure 3 Model AUCs for one-, three-, and five-year CSS. The curves for models (logistic regression (LR), extreme gradient boosting (XGBoost), random forest (RF) regression and external validation based on XGBoost)) in cancer-specific survival (CSS) of non-metastatic colon cancer. The 45-degree straight line represents that the model has similar chances of correctly classifying patients with vs patients without events. AUC=area under the receiver operating characteristic curve.

Table 3 AUCs of Various Predictors in R&M Model

ML		1-Year	3-Year	5-Year
LR	11 predictors	0.68 (0.59–0.79)	0.75 (0.66–0.83)	0.81 (0.72–0.90)
	14 predictors	0.68(0.59–0.77)	0.74 (0.65–0.83)	0.81 (0.71–0.91)
XGBoost	11 predictors	0.71 (0.64–0.79)	0.79 (0.74–0.86)	0.89 (0.82–0.95)
	14 predictors	0.72 (0.64–0.79)	0.80 (0.74–0.88)	0.89 (0.83–0.96)
RF	11 predictors	0.68 (0.61–0.75)	0.78 (0.72–0.85)	0.88 (0.82–0.95)
	14 predictors	0.70 (0.63–0.78)	0.78 (0.72–0.85)	0.89 (0.83–0.95)

Figure 4 Model AUCs for one-, three-, and five-year R&M model with 11 predictors. The curves for models (logistic regression (LR), extreme gradient boosting (XGBoost), random forest (RF) regression) in recurrence and metastasis (R&M) of non-metastatic colon cancer. The 45-degree straight line represents that the model has similar chances of correctly classifying patients with vs patients without events.

Abbreviation: AUC, area under the receiver operating characteristic curve.

Figure 5 Radar plot for importance of predictors in CSS and R&M models with 11 predictors.

Abbreviations: T, tumor; CEA, carcinoembryonic antigen; CSS, cancer-specific survival; R&M, recurrence and metastasis.

Figure 6 Radar plot for importance of predictors in R&M model with 14 predictors.

Abbreviations: T, tumor; CEA, carcinoembryonic antigen; CSS, cancer-specific survival; R&M, recurrence and metastasis; MSI, microsatellite instability.