Classification of large microarray datasets using fast random forest construction.

Random forest is an ensemble classification algorithm. It performs well when most predictive variables are noisy and can be used when the number of variables is much larger than the number of observations. The use of bootstrap samples and restricted subsets of attributes makes it more powerful than simple ensembles of trees. The main advantage of a random forest classifier is its explanatory power: it measures variable importance or impact of each factor on a predicted class label. These characteristics make the algorithm ideal for microarray data. It was shown to build models with high accuracy when tested on high-dimensional microarray datasets. Current implementations of random forest in the machine learning and statistics community, however, limit its usability for mining over large datasets, as they require that the entire dataset remains permanently in memory. We propose a new framework, an optimized implementation of a random forest classifier, which addresses specific properties of microarray data, takes computational complexity of a decision tree algorithm into consideration, and shows excellent computing performance while preserving predictive accuracy. The implementation is based on reducing overlapping computations and eliminating dependency on the size of main memory. The implementation's excellent computational performance makes the algorithm useful for interactive data analyses and data mining.

A novel data-driven prognostic model for staging of colorectal cancer.

BACKGROUND: The aim of this study was to develop a novel prognostic model that captures complex interplay among clinical and histologic factors to predict survival of patients with colorectal cancer after a radical potentially curative resection. STUDY DESIGN: Survival data of 2,505 colon cancer and 2,430 rectal cancer patients undergoing radical colorectal resection between 1969 and 2007 were analyzed by random forest technology. The effect of TNM and non-TNM factors such as histologic grade, lymph node ratio (number positive/number resected), type of operation, neoadjuvant and adjuvant treatment, American Society of Anesthesiologists (ASA) class, and age in staging and prognosis were evaluated. A forest of 1,000 random survival trees was grown using log-rank splitting. Competing risk-adjusted random survival forest methods were used to maximize survival prediction and produce importance measures of the predictor variables. RESULTS: Competing risk-adjusted 5-year survival after resection of colon and rectal cancer was dominated by pT stage (ie, tumor infiltration depth) and lymph node ratio. Increased lymph node ratio was associated with worse survival within the same pT stage for both colon and rectal cancer patients. Whereas survival for colon cancer was affected by ASA grade, the type of resection and neoadjuvant therapy had a strong effect on rectal cancer survival. A similar pattern in predicted survival rates was observed for patients with fewer than 12 lymph nodes examined. Our model suggests that lymph node ratio remains a significant predictor of survival in this group. CONCLUSIONS: A novel data-driven methodology predicts the survival times of patients with colorectal cancer and identifies patterns of cancer characteristics. The methods lead to stage groupings that could redefine the composition of TNM in a simple and orderly way. The higher predictive power of lymph node ratio as compared with traditional pN lymph node stage has specific implications and may address the important question of accuracy of staging in patients when fewer than 12 nodes are identified in the resection specimen.

Clinical outcomes and cost analysis of a "fast track" postoperative care pathway for ileal pouch-anal anastomosis: a case control study.

PURPOSE: Traditional length of hospital stay after ileal pouch-anal anastomosis is 8 to 15 days. Fast track rehabilitation programs reduce stay, but there are concerns that readmission and complication rates may be increased. This study evaluated a fast track pathway after ileoanal pouch surgery. METHODS: One hundred three consecutive patients underwent ileal pouch-anal anastomosis on two colorectal services using a fast track protocol with early ambulation, diet, and defined discharge criteria. Direct hospital costs and 30-day and long-term complication data were collected. Patients were matched to controls managed with traditional care pathways by other colorectal staff. RESULTS: Matching was established for 97 patients. Fast track patients had shorter hospital stay than controls (median 4 vs. 5 days; mean 5.0 vs. 5.9, P = 0.012). Readmission and recurrent operation rates were similar (24 vs. 20 percent, P = 0.49, and 9 vs. 10 percent, P = 0.8, fast track vs. control, respectively). Median direct costs per patient (US$) within 30 days were lower with fast track (5692 vs. 6672, P = 0.001), primarily because of reductions in postoperative management expenses. Complication rates, including pouch failure, bowel obstruction, pouchitis, and anastomotic stricture were comparable. Early discharge (< or = 5 days from surgery) occurred in 79 (77 percent) fast track patients. Failure with early discharge was associated with male gender, reoperations, and anastomotic complications. CONCLUSIONS: Fast track protocol after ileoanal pouch surgery reduces length of stay and hospital costs without increasing complication rates. Successful early discharge usually signals a benign postoperative course.