Sunday, June 20, 2010

Crisis telediagnosis mobile app demo

Technology-driven tools are evolving so that crisis response can be more participatory, and have more advance preparedness. A demo of the Triage4G mobile app (Android and web platforms) occurred at the Clear 4G WiMAX Developers Symposium held at Stanford University, June 15, 2010. The Triage4G app features live video-conferencing and real-time matching between first responders/crisis victims and remote physicians for telediagnosis. (Application download)

Figure 1. Crisis telediagnosis: real-time matching of first responders/victims and remote physicians for live streaming video telediagnosis


Figure 2. Screenshots - a first responder launches the Triage4G app and inputs Triage Tag data. GPS information is use to create case number. A real-time remote physician match is requested.

Figure 3. Screenshots - a physician receives SMS notification of a patient requesting telediagnosis and logs in to the Triage4G app via smartphone or desktop.

Figure 4. Screenshots - when the match is complete, the first responder's screen updates with the case number and a button to launch Qik. The doctor's screen updates to the first responder's Qik page and then the live video steam begins.

Sunday, June 13, 2010

Dollar Van Demos

Easily the phreshest idea from New York’s Internet Week, held in Manhattan June 7-14, was Dollar Van Demos! Dollar Van Demos are a showcase of musicians, rappers, and comedians performing inside a dollar van with real passengers. Numerous Brooklyn-based artists are featured in a collection of videos filmed inside dollar vans as they travel along their usual transportation routes.

The site has new videos from Cocoa Sarai, Tah Phrum Duh Bush, Grey Matter, Zuzuka Poderosa, Kid Lucky, Atlas, BIMB Family, I-John, Top Dolla Raz, Hasan Salaam, illSpokinn, Joya Bravo, Bacardiiiii and Jarel soon!

Figure 1. Example videos from Cocoa Sarai and Top $ Raz

Sunday, June 06, 2010

Rational growth in consumer genomics

The overall tone of the Consumer Genetics Show, held June 2-4, 2010 in Boston MA, was a pragmatic focus on the issues at hand as compared with the enthusiasm and optimism that had marked the conference’s inaugural event last year. One of the biggest shifts was the new programs that some top-tier health service providers have been developing to include genetic testing and interpretation in their organizations. It also became clear that the few widely-agreed upon success stories for genomics in disease diagnosis and drug response (i.e.; warfarin dosing) have been costly to achieve and will not scale to all diseases and all drugs. Cancer continues to be a key killer app for genomics in diagnosis, treatment, prognosis, cancer tumor sequencing, and risk prediction. Appropriate approaches to multigenic risk assessment for health risk and drug response remain untackled. Greater state and federal regulation seems inevitable. Faster-than-Moore’s-law improvements in sequencing costs continue as Illumina dropped the price of whole human genome sequencing for the retail market from $48,000 to $19,500.

There was generally wide agreement that from a public health perspective, personalized genomics is scientifically valid, clinically useful, and reimbursable in specific situations, but not universally – at this time, better information should be obtained for some people, not more information for all people. The focus should be on medical genetics strongly linked to disease, and on pharmacogenomics in treatment. For example, genomic analysis is required for some drugs by the FDA (maraviroc, cetuximab, trastuzumab, and dasatinib), and recommended for several others (warfarin, rasburicase, carbamazepine, abacavir, azathiprine, and irinotecan). A key point is to integrate the drug test with the guidance for drug dosage.

Even when genomic tests are inexpensive enough to be routine, interpretation may be a bottleneck as each individual’s situation is different when taking into account family history, personal medical history, and environmental and other factors. One idea was that the 20,000 pathologists in the US could be a resource for genomic test interpretation; pathologists are already involved as they must certify genetic test data in CLIA labs. Genomic tests and their interpretation would likely need to be standardized and certified in order to be reimbursed in routine medical care. A challenge is that health service payers are not interested or able to drive genomic test product design.

Key science findings

1. The Regulome and Structural Variation

Michael Snyder presented important research that the regulome, the parts of the genome located around the exome (the 1-2% of the genome that codes for protein), may be critical in understanding disease genesis and biological processes. The complexities of RNA are just beginning to be understood. It is known that there is more than just the simple transcription of DNA to RNA involved in controlling gene expression. For example, there is also tight regulation in splicing newly synthesized RNA molecules into the final RNA molecule and in translating messenger RNA to ribosomes to create proteins. Research findings indicate a global/local model of gene regulation, that there are master regulators with universal reach and local regulators operating on a local range of 200 or so genes.

Snyder also presented updates on his lab’s ongoing research into the structural variation of the human genome. A high-resolution sequencing study has been conducted regarding the amount of structural variation in humans, finding that there are ~1,500 structural variations per person that are over 3 kilobases long and that the majority of the structural variations are 3-10 kilobases long with a few extending to 50-100 kilobases (application of this research: Kasowski M, Science, 2010 Apr 9).

2. Reaching beyond the genome to the diseasome, proteome, and microbiome

Several scientists addressed the ways in which science is quickly reaching beyond the single point mutations and structural variation of the genome to other layers of information. There is a need for the digital quantification of the epigenome, the methylome, the transcriptome, the proteome, the metabolome, and the dieaseome/VDJome. For example, the immune system is one of the best monitors of disease state and progression. The strength of individual immune systems can be evaluated through the VDJome (the repertoire of recombined V-D-J regions in immune cells; cumulative immunoglobulin and T-cell receptor antigen exposure)

There are many areas of interest in proteomics including protein profiling, protein-protein interactions, and post-translational modification. A large-scale digital approach to proteomics was presented by Michael Weiner of Affomix. A key focal area is post-translational modification. At least one hundred post-translational modifications have been found, and two are being investigated in particular: phosphorylation (the signal transduction can possibly indicate tumor formation) and glycosylation (possibly indicating tumor progression).

The microbiome (human microbial bacteria) and host-bacteria interactions are an important area for understanding human disease and drug response, and for Procter & Gamble in creating consumer products. The company has basic research and publications underlying products such as the ProX anti-wrinkle skin cream (Hillebrand, Brit Jrl Derm, 2010), rhinovirus, and gingivitis. The company has a substantial vested interest in understanding the microbiome with its variety of nasal, oral, scalp, respiratory, skin, and GI tract-related products.

Sunday, May 30, 2010

Microbubbles and photoacoustic probes energize cancer researchers

The Canary Foundation’s eighth year of activities was marked with a symposium held at Stanford University May 25-27, 2010. The Canary Foundation focuses on the early detection of cancer, specifically lung, ovarian, pancreatic, and prostate cancer, in a three-step process of blood tests, imaging, and targeted treatment.

Imaging advances: microbubbles and photoacoustic probes
Imaging is an area that continues to make advances. One exciting development is the integration of multiple technologies, for example superimposing molecular information onto traditional CT scans. Contemporary scans may show that certain genes are over-expressed in the heart, for example, but obscure the specific nodule (tumor) location. Using integrins to bind to cancerous areas may allow their specific location to be detected (4 mm nodules now, and perhaps 2-3 mm nodules as scanning technologies continue to improve).

Other examples of integrated imaging technologies include microbubbles, which are gassy and can be detected with an ultrasound probe as they are triggered to vibrate. Similarly, photoacoustic probes use light to perturb cancerous tissue, and then sound detection tools transmit the vibrations. Smart probes are being explored to detect a variety of metaloproteases on the surface of cancer cells, breaking apart and entering cancer cells where they can be detected with an ultrasound probe.

Systems biology approaches to cancer
Similar to aging research, some of the most promising progress points in cancer research are due to a more systemic understanding of disease, and the increasing ability to use tools like gene expression analysis to trace processes across time. One example is being able to identify and model not just one, but whole collections of genes that may be expressed differentially in cancers, seeing that whole pathways are disrupted, and the downstream implications of this.

Cancer causality
Also as in aging research, the 'chicken or the egg' problem arises as multiple things that go wrong are identified, but which happens first, and causality, is still unknown. For example, in ovarian cancer, where there are often mutations in the p53 gene, and gene rearrangements and CNV (copy number variation; different numbers of copies of certain genes), but which occurs first and what causes both is unknown.

Predictive disease modeling
There continues to be a need for models that predict clinical outcome, and serve as accurate representations of disease. DNA and gene expression, integrated with traits and other phenotypic data in global coherent datasets could allow the ability to build probabilistic causal models of disease. It also may be appropriate to shift to physics/accelerator-type models to manage the scale of data now being generated and reviewed in biomedicine.

Monday, May 24, 2010

Human-data interface

There is starting to be more data than ever available to individuals, including personal data, every detail of interactions, activities, behaviors, and health habits tracked. Some contend that humans are not ready to interact with massive data and that a common response is to ignore it. However, humans quickly adapt to most situations and this is the likely course with big data interaction.

Data is likely to only become more pervasive and intimate, and extraordinarily useful to those who can harness it.

There is a structural factor with data collection that could be masking the value of the information – timeframes and scale. Data is collected on daily or hourly timeframes, but may be most useful 1) being dormant, a reserve reference with which to compare when anomalies arise, and 2) longitudinally, reflecting attributes over long time scales; seeing a change in sleep patterns over decades for example. Another example is that while it might seem useless to record one's temperature every day, knowing the average temperatures and heart rates of individuals in a community and any deltas (changes) could be quite useful in predicting the spread and magnitude of pandemics such as H1N1.

Sunday, May 16, 2010

Unified health data climate

The future of health management and biosecurity is having always-on access to the health data climate of individuals, families, communities, and countries. A whole new era of health awareness and self-management could be possible. Ideally, health data streams would be automatically captured and parsed into a comprehensive tableau of status monitoring and action-taking.

Key health data streams (Figure 1):

  1. Genome - whole human genome sequence, abnormal tissue sequences (cancer, etc.)
  2. Phenotype - current status of a wide range of biophysical markers including blood-based organ-secreted proteins prognosticating disease, cholesterol levels, blood pressure, and emotional state
  3. Diseasome - catalog of cumulative immune system exposures and predicted response to toxins
  4. Microbiome - microflora bacteria profile (gut, genital, skin, oral, etc.)
  5. Environmentome - external environment measures including air and water quality, pollen/allergens count

Figure 1: Key health data streams.

Sunday, May 09, 2010

The big data graph era

With the start of the big data era and the ability to collect, store and render meaningful numerous data points, the cultural outlook of the world is shifting too. Graphs, graphs, graphs. Individuals and communities have a social graph, taste graph, preference graph, affinity graph, attention graph, intention graph, values graph, emotion graph, health graph and more.

Graphing theory is being applied to many new contexts such as social networks, media consumption, nanotechnology fabrication, gaming, and genomic analysis and could be one of the many data analysis techniques applied to any large dataset. VLDS – very large datasets – and moving back into the cloud mean that sophisticated data analysis and artificial intelligence techniques could be an expected feature of websites just like social networking commentary and gaming elements have become today.

Sunday, May 02, 2010

The preference economy

There is now the new era of a multicurrency society. Numerous non-monetary currencies are coveted, amassed and exchanged including reputation, social graph, time, ideas, intention, attention, affinity, preference, health, and resource access.

The internet is already doing a good job of serving as a clearing exchange and means of valuation for the currencies of reputation, social graph, intention, and attention.

The next generation of economy 3.0 startups is building even more dimensionality into the multicurrency society.

Blippy broadcasts purchasing activity and serves as a leading indicator for public company quarterly sales; a real-time economy feed.
Hunch goes a step further with the grand vision of mapping and predicting the affinity of all people for all objects.
For example, what is any individual’s preference for Nike, TikTok, Slaughterhouse-Five, Ulan Bator, existentialism, or any other noun, brand, product, item, object or idea. Social feed “likes” are already being mined for preference, affinity, and revenue.

Value, preference, and affinity could become an expected attribute of any product, brand, website, and experience just like social networking is and gaming principles are starting to be. These seemingly unobtrusive currencies could stream nicely into exchange via automatic markets.

Sunday, April 25, 2010

Supercomputing and human intelligence

As of November 2009, the world’s fastest supercomputer was the Cray Jaguar located at the U.S. Department of Energy’s Oak Ridge National Laboratory, operating at 1.8 petaflops (1.8 x 1015 flops). Unlike human brain capacity, supercomputing capacity has been growing exponentially. In June 2005, the world’s fastest supercomputer was the IBM Blue Gene/L at Los Alamos National Laboratory, running at 0.1 petaflops. In less than five years, the Jaguar represents an order of magnitude increase, the latest culmination of capacity doublings each few years. (Figure 1)

Figure 1. Growth in supercomputer power
Source: Ray Kurzweil with modifications

The next supercomputing node, one more order of magnitude, at 1016 flops, is expected in 2011 with the Pleiades, Blue Waters, or Japanese RIKEN systems. 1016 flops would possibly allow the functional simulation of the human brain.

Clearly, there are many critical differences between the human brain and supercomputers. Supercomputers tend to be modular in architecture and address specific problems as opposed to having the general problem solving capabilities of the human brain. Having equal to or greater than human-level raw computing power in a machine does not necessarily confer the ability to compute as a human. Some estimates of the raw computational power of the human brain range between 1013 and 1016 operations per second. This would indicate that
supercomputing power is already on the order of estimated human brain capacity, but intelligent or human-simulating machines do not yet exist.
The digital comparison of raw computational capability may not be the right measure for understanding the complexity of the brain. Signal transmission is different in biological systems, with a variety of parameters such as context and continuum determining the quality and quantity of signals.

Sunday, April 18, 2010

Radical transparency

Social networking and Web 2.0 has made it easy to find out about the friends, resume, activities, and interests of the many people who permission-in and broadcast this information.

Financial privacy disappeared for groups of the population as benefits outweighed costs in peer-to-peer lending, real estate, expense management, and purchasing with Prosper, Zillow, Expensr, Mint, and now Blippy.

Health data is the new frontier as people are starting to publicly post their genome files, and perhaps blood test information with the SNPedia, Personal Genome Project, and DIYgenomics. Some people are tweeting their weight, and could possibly do so with their sleep-tracking Z scores and other quantified self tracking activities.

In the farther future, who will be the first to tweet their neural feed? The unexpurgated feed that would be captured directly from the brain, not medicated by language, typing, consciousness, and culture as now. As with other successful technology roll-out paradigms, truth culture is likely to be opt-in, and the competitive advantage could likely be with those who do decide to disclose.