AAAI 2014: Connecting Machine Learning and Human Intelligence

The AAAI Spring Symposia are a place for worldwide artificial intelligence, machine learning, and other computer scientists to present and discuss innovative theoretical research in a workshop-like environment. In 2013, some of the topics included: learning, autonomous systems, wellness, crowd computing, behavior change, and creativity.

Proposals are underway for 2014. Please indicate your opinion by voting at the poll at the top right for these potential topics:

• My data identity: personal, social, universal 
• Big data becomes personal: knowledge into meaning 
• Wearable computing and digital affect: wellness, bioidentity, and intentionality 
• Big data, wearable computing, and identity construction: knowledge becomes meaning 
• Personalized knowledge generation: identity, intentionality, action, and wellness

Key challenge of our era: health and preventive medicine

Delivering health care and keeping populations healthy is a key problem of the current era. Health expenditures currently comprise 17% of U.S. GDP and are growing; simultaneously health in the U.S. is in decline, with a new CDC report estimating that by 2030, 42% of American adults will be obese, compared to 34% today and 11% will be severely obese, compared to 6% today.

The Realization of Preventive Medicine
A key part of addressing health challenges is the realization of preventive medicine. Preventive medicine and health maintenance consist of identifying and managing conditions in the 80% of their life cycle before they become clinical, ideally avoiding clinical onset. Workable models for the execution of preventive medicine need to be developed. By definition, a broader ecosystem than the traditional medical establishment will be participating in all steps of the value chain ranging from health research to clinical delivery. More flexible regulatory models are needed that preserve the core ethical principles of the traditional models, but are geared towards the internet era and an expanded notion of health and health maintenance with a larger ecosystem of service providers and participants. The payments ecosystem needs to adapt in parallel, allowing for a wider range of payment mechanisms including out-of-pocket payments, H.S.A. dollars, patient advocacy group funding, and traditional (and increasingly diminishing) insurance payments.

Health 2.0 business models

Health 2.0 is about re-envisioning every aspect of health and health care. New business models are starting to develop to support this innovation ecology. First, accompanying the new paradigm of community research (peer cohort studies à la Patients Like Me (lithium) and DIYgenomics (MTHFR mutation/Vitamin B-12 deficiency), could be social venture finance, corporate sponsorship from supplement companies and other remedy vendors, crowdsourced finance (i.e.; Kickstarter), and philanthropist contributions. Second, the traditional venture capital model is already being applied to health 2.o startup companies, including through organizations such as the Health 2.0 Accelerator. Third, whole new industries may sprout from the nascent efforts of health advisors and wellness coaches. The health advisor is the analog to the financial advisor or mortgage broker, able to integrate a client's health data streams, needs, and interests with available offerings, across a spectrum of economic models: insurance reimbursable, HSA dollars, and direct out-of-pocket spending.

Individuals to drive personalized medicine era

The Personalized Medicine World Congress held January 19-20, 2010 in Mountain View, CA was one of the first business conferences devoted to personalized medicine. There is a lot of excitement about personalized medicine and genomics given some recent announcements regarding whole human genome sequencing. First, Complete Genomics reported the costs of consumables (required chemical reagents), dropping to $4,400, and even $1,500 (Supporting Online Material page 27) per genome. Illumina similarly announced dramatic price drops, an estimated all-in cost of $10,000 per whole human genome with the new HiSeq 2000 machine. Illumina currently charges individuals $48,000 for whole human genome sequencing. The HiSeq 2000 is priced at $690,000 per machine and BGI (formerly the Beijing Genome Institute) has ordered 128.

Complete Genomics’ CEO Cliff Reid made an interesting point that despite genomic sequencing having been progressing at 10x improvements per year since 2006, theoretical limits are starting to be reached and the industry will probably return to regular Moore’s Law progress curves (18 month performance doublings). While third-generation sequencers such as Complete Genomics (using a short-read sequencing-by-probe-ligation technology) and Pacific Biosciences (using a single-molecule real-time sequencing by synthesis technology) may start to reach limits, fourth-generation sequencers using other technologies such as nanopores (e.g., Oxford Nanopore Technologies), and electron microscope imaging (e.g., Halcyon Molecular, ZS Genetics), may be able to keep the sequencing industry progressing at faster-than-Moore’s-Law rates.

The most hopeful comments came from Esther Dyson and Leroy Hood. Esther Dyson, pointing out the still heavy focus on health institutions rather than consumer-empowerment for transformation to the personalized medicine era said that she felt like she was “representing the PC world at a mainframe convention.” Directly paralleling the current medical system, she also noted that when Gutenberg arrived with the printing press, the priests said ‘there’s no reason people need to read the bible themselves, we can read it for them.’ However, as shown in Figure 1, personalized medicine is about wellness, not disease, and while there are certainly overlaps with the current domain of physicians, there may be minimal encroachment due to automated tools and new health ecosystem participants such as wellness advisors.

Figure 1. Wellness becomes the domain and responsibility of the individual

Leroy Hood set forth a detailed plan for the future of medicine, P4 Medicine: medicine that is predictive, personalized, preventive, and participatory. Looking for the fingerprints of health vs. disease, he envisions a future where billions of data points are investigated per individual. There could be at least four relevant data sets. One data set is the whole human genome sequence. Another could be a biannual wellness screen for 2,500 blood-based organ-specific proteins indicating possible precursors to disease. A third data set could be an immune system screen of the 10,000 B cells and 10,000 T cells, looking at the functional regions of immune receptors, and past and preset immune responsiveness. A fourth data set, in the instance of cancer, could be taking a single cancer call and sequencing 1,000 transcriptomes simultaneously to understand how cancer is expressed in particular individuals. These data sets could help to realize medicine as an information science and address the specificity of disease and wellness in individuals.