ABSTRACT
Contents Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.1.1 Seven important papers (out of so many more) . . . . . . . . 3 1.1.1.1 Study design. . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 1.1.1.2 Database design . . . . . . . . . . . . . . . . . . . . . . . . . 5 1.1.1.3 Sampling and screening. . . . . . . . . . . . . . . . . . . 5 1.1.1.4 Randomization. . . . . . . . . . . . . . . . . . . . . . . . . . 6 1.1.1.5 Data entry and monitoring. . . . . . . . . . . . . . . . . 6 1.1.1.6 Data analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 1.1.1.7 Publication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 1.1.1.8 Data sharing. . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 1.1.1.9 Reproduction of central ndings . . . . . . . . . . . . 7 1.1.1.10 Planning the next study . . . . . . . . . . . . . . . . . . . 7
1.2 Future of autism data science . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 1.2.1 Data harmonization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.2.1.1 Some pros and cons of meta-analysis . . . . . . . . 8 1.2.1.2 Planning multisite studies . . . . . . . . . . . . . . . . . 9
1.2.2 Data display and its diagnostic limitations. . . . . . . . . . . . 9 1.2.2.1 Is it possible to diagnose autism by brain
imaging?. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 1.2.3 Candidate phenotypes. . . . . . . . . . . . . . . . . . . . . . . . . . . 11 1.2.4 Reporting results. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
1.3 Concluding remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
Abstract This chapter begins with a critical look at the present state of autism data science in observational studies and clinical trials. It then proceeds to list the necessary steps in the proper application of autism data science for young researchers, and experienced researchers who may have seen this previously; it is worth taking another look. Multisite studies and cluster randomization methods are also discussed. It includes a section on the future state of autism imaging and its limits at present. The chapter concludes with some views and speculations on future improvement of autism data science.
