Reporter: Larry H Bernstein, MD, FCAP
Pathologists May Be Healthcare’s Rock Stars of Big Data in Genomic Medicine’s ’Third Wave’
Pathologists are positioned to be the primary interpreters of big data as genomic medicine further evolves
Pathologists and clinical laboratory managers may be surprised to learn that at least one data scientist has proclaimed pathologists the real big data rock stars of healthcare. The reason has to do with the shift in focus of genomic medicine from therapeutics and presymptomatic disease assessment to big data analytics.
In a recent posting published at Forbes.com, data scientist Jim Golden heralded the pronouncement of Harvard pathologist Mark S. Boguski, M.D., Ph.D., FACM. He declared that “The time of the $1,000 genome meme is over!”
DNA Sequencing Systems and the $1,000 Genome
Golden has designed, built, and programmed DNA sequencing devices. He apprenticed under the Human Genome program and spent 15 years working towards the $1,000 genome. “I’m a believer,” he blogged. “[That’s] why I was so intrigued [by Boguski’s remarks].
Boguski is Associate Professor of Pathology at the Center for Biomedical Informatics at Harvard Medical School and the Department of Pathology at Beth Israel Deaconess Medical Center. It was in a presentation at a healthcare conference in Boston that Boguski pronounced that it is time for the $1,000 genome to go.
Both Golden and Boguski acknowledged the benefits generated by the race to the $1,000 genome. Competition to be first to achieve this milestone motivated scientists and engineers to swiftly drive down the cost of decoding DNA. The result was a series of advances in instrumentation, chemistry, and biology.
Pathologists and Big Data Analytics
“Our notions about how genome science and technology would improve health and healthcare have changed,” Boguski wrote in an editorial published at Future Medicine. He then noted that the focus has shifted to big data analytics.
In the editorial, Boguski described the phases of development of genomic medicine as “waves.” The first wave occurred during the mid- to late-1990s. It focused on single- nucleotide polymorphisms (SNP) and therapeutics.
Medical Laboratories Have Opportunity to Perform Presymptomatic Testing
The second wave focused on presymptomatic testing for disease risk assessment and Genome Wide Association Studies (GWAS). Researchers expected this data to help manage common diseases.
The first two waves of medical genomics were conducted largely by the pharmaceutical industry, as well as with primary care and public health communities, according to Boguski. Considerable optimism accompanied each wave of medical genomics.
“Despite the earlier optimism, progress in improving human health has been modest and incremental, rather than paradigm-shifting,” noted Boguski, who wrote that,to date, only a handful of genome-derived drugs have reached the market. He further observed that products such as direct-to-consumer genomic testing have proved more educational and recreational than medical.
“Third Wave” of Genomic Medicine
It was rapid declines in the cost of next-generation DNA sequencing technologies that now has triggered the third wave of genomic medicine. Its focus is postsymptomatic genotyping for individualized and optimized disease management.
“This is where genomics is likely to bring the most direct and sustained impact on healthcare for several reasons,” stated Boguski. “Genomics technologies enable disease diagnosis of sufficient precision to drive both cost-effective [patient] management and better patient outcomes. Thus, they are an essential part of the prescription for disruptive healthcare reform.”
Boguski reiterated the case for the value of laboratory medicine. He stated the following critical—but often overlooked—points, each of which is familiar to pathologists and clinical laboratory managers:
1. Pathologist-directed, licensed clinical laboratory testing has a major effect on clinical decision-making.
2. Medical laboratory testing services account for only about 2% of healthcare expenditures in the United States.
3. Medical laboratory services strongly influence the remaining 98% of costs through the information they provide on the prevention, diagnosis, treatment, and management of disease.
Molecular Diagnostics Reaching Maturity for Clinical Laboratory Testing
“Genome analytics are just another technology in the evolution of molecular diagnostics,” Boguski declared in his editorial.
Read more: Pathologists May Be Healthcare’s Rock Stars of Big Data in Genomic Medicine’s ’Third Wave’ | Dark Daily http://www.darkdaily.com/pathologists-may-be-healthcare%e2%80%99s-rock-stars-of-big-data-in-genomic-medicines-third-wave-1217#ixzz2FL24IRAA
- Digital Pathology Market Expected to Reach $336.6 Million at a CAGR of 12% from 2012 to 2017 by MarketsandMarkets (prweb.com)
- Response to UK’s Plan to sequence genomes of 100,000 Cancer Patients (grosenfeld.wordpress.com)
- The Fast Changing Landscape of Sequencing Technologies and Their Impact on Microbial Genome Assemblies and Annotation (plosone.org)
- Stanford to rival Broad Institute with Big Data-focused genomics center (fiercebiotechit.com)
- New Method To Directly Sequence Small Genomes Without Library Preparation (medicalnewstoday.com)
- Three Big Themes from the Personalized Medicine Conference (xconomy.com)
- NIH Grant Moves Pathologists to the Forefront of Genomic Medicine (prweb.com)
- Tool developed to evaluate genome sequencing method (rdmag.com)
- Big Data To Become Powerful Driver Of IT Spending – Gartner (misco.co.uk)
- Video: DDN Powers Genomics at the Virginia Bioinformatics Institute (insidehpc.com)
Comment & Response
Right now the cost of the testing and the turnaround times are not favorable. It is going to take a decade or more for clinical labs to catch up. For some time it will be send out tests to Quest, LabCorp, and State or University lab consortia.
The power of the research technology is pushing this along, but for Personalized Medicine the testing should be coincident with the patient visit, and the best list of probable issues should be accessible on the report screen. The EHR industry is dominated by 2 companies that I see have no interest in meeting the needs of the physicians. The payback has to be on efficient workflow, accurate assessment of the record, and timely information. The focus for 25 years has been on billing structure. But even the revised billing codes (ICD10) can’t be less than 5 years out-of-date because of improvements in the knowledge base and improvements in applied math algorithms.
The medical record still may have information buried min a word heap, and the laboratory work is a go-to-you know where sheet with perhaps 15 variables on a page, with chemistry and hematology, immunology, blood bank, and microbiology on separate pages. The ability of the physician to fully digest the information with “errorless” discrimination is tested, and the stress imposed by the time for each patient compromises performance. There is work going on in moving proteomics along to a high throughput system for improved commercial viability, that was reported by Leigh Anderson a few years ago. The genomics is more difficult, but the genomics is partly moving to rapid micropanel tools.
In summary, there are 3 factors:
1. Automation and interpretation
2. Integration into the EHR in real time and usable by a physician.
3. The sorting out of the highest feature “predictors” and classifying them into clinically meaningful sets and subsets.
When this is done, then the next generation of recoding will be in demand.
The Automated Malnutrition Assessment
Gil David1, Larry Bernstein2, Ronald R. Coifman1
1Department of Mathematics, Program in Applied Mathematics,
Yale University, New Haven, CT 06510, USA,
2Triplex Consulting, Trumbull, CT 06611
Introduction: We propose an automated nutritional assessment (ANA) algorithm that provides a method for malnutrition risk prediction with high accuracy and reliability.
Materials and Methods: The database used for this study is a file of 432 patients, where each patient is described by 4 laboratory parameters and 11 clinical parameters. A malnutrition risk assessment of low (1), moderate (2) or high (3) was assigned by a dietitian for each patient. An algorithm for data organization and classification via characteristic metrics is proposed. For each patient, the algorithm characterizes its unique profile and builds a characteristic metric to identify similar patients who are mapped into a classification.
Results: The algorithm assigned a malnutrition risk level for each patient based on different training sizes that were taken out of the data.
Our method resulted in an average error (distance between the automated score and the real score) of 0.386, 0.3507, 0.3454, 0.34 and 0.2907 for 10%, 30%, 50%, 70% and 90% training sizes, respectively.
Our method outperformed the compared method even when our method used a smaller training set then the compared method. In addition, we show that the laboratory parameters themselves are sufficient for the automated risk prediction and organize the patients into clusters that correspond to low, low-moderate, moderate, moderate-high and high risk areas.
Discussion: The problem of rapidly identifying risk and severity of malnutrition is crucial for minimizing medical and surgical complications. These are not easily performed or adequately expedited. We characterize for each patient a unique profile and map similar patients into a classification. We also find that the laboratory parameters themselves are sufficient for the automated risk prediction.
Keywords: Network Algorithm, unsupervised classification, malnutrition screening, protein energy malnutrition (PEM), malnutrition risk, characteristic metric, characteristic profile, data characterization, non-linear differential diagnosis.