Context

There's a story behind every dataset and here's your opportunity to share yours.

Content

What's inside is more than just rows and columns. Make it easy for others to get started by describing how you acquired the data and what time period it represents, too. Welcome to our regular closed dataset based weekend hackathon. In this weekend hackathon, we are challenging all the machinehackers to predict the melanoma tumor size based on various attributes. Melanomas present in many different shapes, sizes, and colors. That’s why it’s tricky to provide a comprehensive set of warning signs. Melanoma, also known as malignant melanoma, is a type of skin cancer that develops from the pigment-producing cells known as melanocytes. The primary cause of melanoma is ultraviolet light (UV) exposure in those with low levels of the skin pigment melanin. The UV light may be from the sun or other sources, such as tanning devices.

Melanoma is the most dangerous type of skin cancer. Globally, in 2012, it newly occurred in 232,000 people. In 2015, there were 3.1 million people with active disease, which resulted in 59,800 deaths. Australia and New Zealand have the highest rates of melanoma in the world. There are also high rates in Northern Europe and North America, while it is less common in Asia, Africa, and Latin America. In the United States melanoma occurs about 1.6 times more often in men than women.

Data Description:

Train.csv - 9146 rows x 9 columns
Test.csv - 36584 rows x 8 columns
Sample Submission - Acceptable submission format 

Attributes Description:

mass_npea: the mass of the area understudy for melanoma tumor
size_npear: the size of the area understudy for melanoma tumor
malign_ratio: ration of normal to malign surface understudy
damage_size: unrecoverable area of skin damaged by the tumor
exposed_area: total area exposed to the tumor
std_dev_malign: standard deviation of malign skin measurements
err_malign: error in malign skin measurements
malign_penalty: penalty applied due to measurement error in the lab
damage_ratio: the ratio of damage to total spread on the skin
tumor_size: size of melanoma_tumor

Acknowledgements

We wouldn't be here without the help of others. If you owe any attributions or thanks, include them here along with any citations of past research.

Inspiration

Your data will be in front of the world's largest data science community. What questions do you want to see answered?

While gender strongly influences survival and presentation in melanoma, little is known about impact of gender in American Indians and Alaskan Natives (AI/AN). This study explored differences in tumor characteristics and survival between AI/AN males and females with invasive melanoma. Using the 2004-2018 National Cancer Database, a retrospective cohort study of AI/AN with primary invasive cutaneous melanoma was conducted. Statistical analysis included Mann-Whitney U (continuous variables), Chi-squared (categorical variables), Kaplan-Meier and log rank test (overall survival (OS)), and a multivariate Cox regression (independent survival predictors). Among AI/AN with invasive melanoma, women are diagnosed at an earlier age and stage, and they have better OS than men. Male gender is an independent predictor of worse OS. This dataset provides supplemental methods that were not able to be described in the research letter due to the limited word count. Additionally, we provide a figure that provides a detailed description of the case selection for this cohort of patients.

Context

Welcome to our regular closed dataset based weekend hackathon. In this weekend hackathon, we are challenging all the machinehackers to predict the melanoma tumor size based on various attributes. Melanomas present in many different shapes, sizes, and colors. That’s why it’s tricky to provide a comprehensive set of warning signs. Melanoma, also known as malignant melanoma, is a type of skin cancer that develops from the pigment-producing cells known as melanocytes. The primary cause of melanoma is ultraviolet light (UV) exposure in those with low levels of the skin pigment melanin. The UV light may be from the sun or other sources, such as tanning devices.

Melanoma is the most dangerous type of skin cancer. Globally, in 2012, it newly occurred in 232,000 people. In 2015, there were 3.1 million people with active disease, which resulted in 59,800 deaths. Australia and New Zealand have the highest rates of melanoma in the world. There are also high rates in Northern Europe and North America, while it is less common in Asia, Africa, and Latin America. In the United States melanoma occurs about 1.6 times more often in men than women.

Content

Train.csv - 9146 rows x 9 columns Test.csv - 36584 rows x 8 columns Sample Submission - Acceptable submission format

Attributes

Acknowledgements

Machine Hack: Melanoma Tumor Size Prediction

Since 2002, the Interdisciplinary Melanoma Cooperative Group (IMCG) at Perlmutter Cancer Center has maintained one of the largest clinicopathologic resources, the Melanoma Clinicopathological-Biospecimen Database and Repository, for research on patients 18 years old and over with melanoma or at high risk for melanoma. Clinical data is stored in a secure REDCap database which contains 653 fields to capture clinical and pathological information. The database can be queried for research studies; customized datasets for statistical analyses are created in SAS®. Follow-up data is collected every 3, 6, or 12 months depending on the patient's clinical stage. Biospecimens (i.e., blood/buffy coat, sera, plasma, lymphocytes; and blocks of primary, metastatic, and fresh melanoma tissues) are securely cataloged in LabVantage with linkage to corresponding clinical and pathological data contained in REDCap. Integration of high-quality, annotated biospecimens with clinicopathological data allow applications such as the examination of RNA expression (fresh tissue), protein expression (paraffin embedded tissue), and germline DNA sequences (blood) from the same patients.

As of March 2023, 5,790 consenting patients (including 399 high-risk patients) have contributed clinical data and 99,039 biospecimens to the project. 2,977(55%) of patients are male; the mean age at diagnosis was 60 years old with a mean follow-up duration of 55 months. These metrics are subject to change over time.

Prioritization Plan for Biospecimen Distribution

To use the resources in the Melanoma Clinicopathological-Biospecimen Database and Repository, investigators need to fill the attached request form. The request is reviewed by the IMCG Biospecimen Committee, consisting of:

• Iman Osman, MD – Director, IMCG
• Andre Moreira, MD, PhD – Director of NYU CBRD
• Yongzhao Shao, PhD – Director, Biostatistics & Bioinformatics
• Richard Shapiro, MD – Director of Surgical Oncology Operations, Surgical Oncology
• Amanda Lund PhD. Associate professor of Dermatology and Pathology

The Committee meets monthly to make decisions regarding distribution of biospecimens based on the scientific merit and status of funding, with priority given to investigators with peer-reviewed funding for projects requiring evaluation of specific biospecimens. Prioritization will be as follows:

1. NYU Melanoma SPORE research projects
2. NYU Melanoma SPORE CEP and DRP projects
3. Inter-SPORE projects
4. Other NCI, NIH, DOD, federally funded or American Cancer Society peer-reviewed projects
5. Non-NCI, NIH, DOD, federally funded or American Cancer Society peer-reviewed projects
6. Non-peer reviewed, Industry-sponsored or no funding

If a conflict arises between two (or more) competing interests within the same category (e.g., two SPORE research projects), the committee decides based on the following criteria:

• Amount of tissue (or specimen) available
• Nature of the specimens (primary versus metastases)
• Specific histologic subtype (e.g., acral-focused projects)
• Site specific metastases (e.g., brain met–focused projects)
• How much material is needed for each project
• Availability of the material (e.g., FFPE specimens are more readily available than fresh or frozen tissues)
• Importance of this specific specimen to the project (e.g., 1 specimen of 50 or 1 of 200 needed)
• Necessity of follow-up clinical information linked to the specimen versus only baseline characteristics

For any project that potentially requires prospective collection, the Biospecimen Committee will attempt to acquire enough materials to allow multi-investigator utilization.

Supplementary Table S2 from Genomic Features of Exceptional Response in Vemurafenib ± Cobimetinib–treated Patients with BRAFV600-mutated Metastatic Melanoma

BackgroundImmunotherapy agents are approved for adjuvant treatment of stage III melanoma; however, evidence for survival benefit in early stage III disease is lacking. Current guidelines for adjuvant immunotherapy utilization in stage IIIA rely on clinician judgment, creating an opportunity for significant variation in prescribing patterns. This study aimed to characterize current immunotherapy practice variations and to compare patient outcomes for different prescribing practices in stage IIIA melanoma.Study designPatients with melanoma diagnosed from 2015-2019 that met American Joint Committee on Cancer 8th edition criteria for stage IIIA and underwent resection were identified in the National Cancer Database. Multiple imputation by chained equations replaced missing values. Factors associated with receipt of adjuvant immunotherapy were identified. Multivariable Cox proportional hazards regression compared overall survival across groups.ResultsOf 4,432 patients included in the study, 34% received adjuvant immunotherapy. Patients had lower risk-adjusted odds of receiving immunotherapy if they were treated at an academic center (OR=0.48, 95%CI=0.33-0.72, p

Supplementary Table S7 from Genomic Features of Exceptional Response in Vemurafenib ± Cobimetinib–treated Patients with BRAFV600-mutated Metastatic Melanoma

RPPA analysis in WM46 cells treated with testosterone