Enterprise Architecture

Precision Medicine: With new program from White House; also comes redundant grant funding and waste – How does all these escape in high science areas?

Slide2

Recently announced Precision Medicine a fantastic mission to bring all the research institutions country wide to collaborate together and holistically solve the civilization’s most complex and pressing problem Cancer, employing genomics while engaging science in an integrative discipline approach.

While the Precision Medicine mission is grand and certainly requires much attention and focus; that many new tools are now available for medical research such as complex algorithms in the areas of cognitive science (data mining, deep learning, etc), bigdata processing, cloud computing, etc; we also need efforts to arrest redundant spend and grants.  

Speaking of precision medicine such waste what an irony.

The White House Hosts a Precision Medicine Initiative Summit

Grand Initiative Redundant Research Grants for Same Methods

$1,399,997 :- Study Description: We propose to develop Bayesian double-robust causal inference methods that are accurate, vigorous, and efficient for evaluating the clinical effectiveness of ATSs, utilizing electronic health records and registry studies, through working closely with our stakeholder advisory panel. The proposed “PCATS” R package will allow easy application of our methods without requiring R programming skills. We will assess clinical effectiveness of the expert-recommended ATSs for the pJIA patient population using a multicenter new-patient registry study design. The study outcomes are clinical responses and the health-related quality of life after a year of treatment.

$832,703 :- Bayesian statistical approach in contrary try to use present as well as historical trial data in a combined framework and can provide better precision for CER. Bayesian methods also flexible in capturing subjecting prior opinion about multiple treatment options and tend to be robust. Despite these advantages, the Bayesian method for CER is underused and underdeveloped (see PCORI Methodology Report, pg. 64, 2013). The primary reasons being a lack of understanding about the role, the lack of methodological development, and the unavailability of easy-to-use software to design and conduct such analysis.

$839,943 :- We propose to use a method of analysis called Bayes method, in which data on the frequency of a disease in a population is combined with data taken from an individual patient (for example, the result of a diagnostic test) to calculate the chance that the patient has the disease given his or her test result. Clinicians currently use Bayes method when screening patients for disease, but we believe the utility of this methodology extends far beyond its current use.

$535,277 Specific Aims:

  1. To encourage Bayesian analysis of HTE:
  • To develop recommendations on how to study HTE using Bayesian statistical models
  • To develop a user-friendly, free, validated software for Bayesian methods for HTE analysis

2. To develop recommendations about the choice of treatment effect scale for the assessment of HTE in PCOR. The main products of this study will be:

  • recommendations or guidance on how to do Bayesian analysis of HTE in PCOR
  • software to do the Bayesian methods
  • recommendations or guidance on choosing appropriate treatment effect scale for HTE analysis in PCOR, and
  • demonstration of our products using data from large comparative effectiveness trials.
Advertisements

Bioingine.com; Integrated Platform for Population Health and EBM based Patient Health Analytics

Ingine Inc; Bioingine.com

Deductive Logic to Inductive Logic

Notational Algebra & Symbolic Programming

Deductive – What | Inductive – Why, How

Deductive:- Statistical Summary of the Population by each Variable Recorded

Deductive:- Statistical Distribution of a Variable

Deductive:- Partitioning Data into Clusters

Cluster analysis is an unsupervised learning technique used for classification of data. Data elements are partitioned into groups called clusters that represent proximate collections of data elements based on a distance or dissimilarity function. Identical element pairs have zero distance or dissimilarity, and all others have positive distance or dissimilarity.

http://www.francescobonchi.com/CCtuto_kdd14.pdf

 

correlation coefficient is a coefficient that illustrates a quantitative measure of some type of correlation and dependence, meaning statistical relationships between two or more random variables or observed data values.

The regression equation can be thought of as a mathematical model for a relationship between the two variables. The natural question is how good is the model, how good is the fit. That is where r comes in, the correlation coefficient (technically Pearson’s correlation coefficient for linear regression).

Inductive :- Hyperbolic Dirac Net 

Notes on Synthesis of Forms :-

Christopher Alexander on Inductive Logic

The search for causal relations of this sort cannot be mechanically experimental or statistical; it requires interpretation: to practice it we must adopt the same kind of common sense that we have to make use of all the time in the inductive part of science. The data of scientific method never go further than to display regularities. We put structure into them only by inference and interpretation. In just the same way, the structural facts about a system of variables in an ensemble will come only from the thoughtful interpretation of observations.

We shall say that two variables interact if and only if the designer can find some reason (or conceptual model) which makes sense to him and tells him why they should do so.

But, in speaking of logic, we do not need to be concerned with processes of inference at all. While it is true that a great deal of what is generally understood to be logic is concerned with deduction, logic, in the widest sense, refers to something far more general . It is concerned with the form of abstract structures, and is involved the moment we make pictures of reality and then seek to manipulate these pictures so that we may look further into the reality itself . It is the business of logic to invent purely artificial structures of elements and relations. 

Christopher Alexander:- Sometimes one of these structures is close enough to a real situation to be allowed to represent it. And then, because the logic is so tightly drawn, we gain insight into the reality which was previously withheld from us.

Quantum Mechanics Driven Knowledge Inference for Medical Diagnosis

http://www.bioingine.com/?p=528

HDN Inference

HDN Results :- Inverse Bayesian Probability

(more…)

Platform for BigData Driven Medicine and Public Health Studies [ Deep Learning & Biostatistics ]

Panel_Logo

Bioingine.com; Platform for comprehensive statistical and probability studies for BigData Driven Medicine and Public Health.

Importantly helps redefine Data driven Medicine as:-

Ontology (Semantics) Driven Medicine

Comprehensive Platform that covers Descriptive Statistics and Inferential Probabilities.

Beta Platform on the anvil. Signup for Demo by sending mail to

“demo@bioingine.com”

Bioingine.com employs algorithmic approach based on Hyperbolic Dirac Net that allows inference nets that are a general graph (GC), including cyclic paths, thus surpassing the limitation in the Bayes Net that is traditionally a Directed Acyclic Graph (DAG) by definition. The Bioingine.com approach thus more fundamentally reflects the nature of probabilistic knowledge in the real world, which has the potential for taking account of the interaction between all things without limitation, and ironically this more explicitly makes use of Bayes rule far more than does a Bayes Net.

It also allows more elaborate relationships than mere conditional dependencies, as a probabilistic semantics analogous to natural human language but with a more detailed sense of probability. To identify the things and their relationships that are important and provide the required probabilities, the Bioingine.com scouts the large complex data of both structured and also information of unstructured textual character.

It treats initial raw extracted knowledge rather in the manner of potentially erroneous or ambiguous prior knowledge, and validated and curated knowledge as posterior knowledge, and enables the refinement of knowledge extracted from authoritative scientific texts into an intuitive canonical “deep structure” mental-algebraic form that the Bioingine.com can more readily manipulate.

BigData Driven Medicine Program :-

http://med.stanford.edu/iddm.html

Objectives and Goals

Informatics & Data-Driven Medicine (IDDM) is a foundation area within the Scholarly Concentration program that explores the new transformative paradigm called BIG DATA that is revolutionizing medicine. The proliferation of huge databases of clinical, imaging, and molecular data are driving new biomedical discoveries and informing and enabling precision medical care. The IDDM Scholarly Concentration will provide students insights into this important emerging area of medicine, and introducing fundamental topics such as information management, computational methods of structuring and analyzing biomedical data, and large-scale data analysis along the biomedical research pipeline, from the analysis and interpretation of new biological datasets to the integration and management of this information in the context of clinical care.

Requirements

Students who pursue Informatics & Data-Driven Medicine in conjunction with an application area, such as Immunology, are required to complete 6 units including:

Biomedin 205: Precision Practice with Big Data

Bioingine.com :- Quantum Mechanics Machinery for Healthcare Ecosystem Analytics

Screenshot 2016-04-01 10.25.05

Notational – Symbolic Programming Introduced for Healthcare Analytics

Quantum Mechanics Firepower for Healthcare Ecosystem Studies        

Interoperability Analytics

Public Health and Patient Health

Quantum Mechanics Driven A.I Experience

Deep Machine Learning

Descriptive and Inferential Statistics

Definite and Probabilistic Reasoning and Cognitive Experience

Know Your Health Ecosystem (Semantic Lake) :- Deep Learning from Healthcare Interoperability BigData – Descriptive and Inferential Statistics

Screenshot 2016-04-01 11.28.31

Bioingine.com; Platform for Healthcare Interoperability (large data sets) Analytics

Deep Learning from Millions of EHR Records

1. Payer – Provider:- (Mostly Descriptive Statistics)

Mostly answers “What”

  • Healthcare Management Analysis (Systemic Efficiencies)
  • Opportunities for cost reduction
  • Chronic patient management
  • Pathway analysis for cost insights
  • Service based to Performance Based – Outcome Analysis (+Inferential)

2. Provider – Clinical Data – (Mostly Inferential Statistics)

Reasoning to understand “Why”, “How”, “Where” (Spatial) and “When” (Temporal)

  • Healthcare Delivery Analysis (Clinical Efficacies)
  • EBM – Clinical Decision Support – Hypothesis Analysis
  • Pathways and Outcome (+Descriptive)

Health Information Exchange :- Interoperability Large BigData

HDN_Cognitive_Computing

Sample Descriptive Statistics:-

Inferential Statistics:-

Computing Cardiovascular Risk Score – BioIngine (HDN)

Mar 21, 2016

Hyperbolic Dirac Net :- Dual Bayesian Probabilities Computations.

Below is Screen Shots of the BioIngine – HDN Ingine; that creates a semantic lake and from which computes the probabilistic risk; as one of the use case.

Bioingine.com; in its first attempt is being implemented employing the Wolfram Symbolic Programming Language, where the Bioingine’s algorithm is developed based on Q-UEL (a Notional Language Framework to express Probabilistic Semantics – derived from the Dirac Notation a Framework to describe Quantum Mechanics). As such the Wolfram Language is designed based on deterministic system providing definite answers based on deductive methods.

Introduction of Q-UEL / HDN introduces probabilistic system, where knowledge by inference can be extracted employing inductive techniques.

A typical query below; which is a Dirac Notational Expression (semantic algebra) queries the semantic lake (the health ecosystem as HDN representation)

HDN Inference Results

Pay attention to notational expression and HDN computations in Section B.

Section A
‘HEART ATTACK’ | ‘Age(years)’:=’50-59′> ‘HEART ATTACK’ | ‘History of BP’:=’1′ and ‘Taking BP medication’:=’1′ and ‘History of high cholesterol’:=’1′>/’HEART ATTACK’ | ? >’History of BP’:=’1′ and ‘Taking BP medication’:=’1′ | ‘Total cholesterol(mg/dl)’:=’180-229′> ‘History of high cholesterol’:=’1′ | ‘Male’:=’0′ and ‘Systolic BP(mmHg)’:=’170-179′ and ‘Taking diabetes medication’:=’0′>  /’History of BP’:=’1′ | ? > /’Taking BP medication’:=’1′ | ? >/’History of high cholesterol’:=’1′ | ? >

Got Following Results:- Pay attention to how all the probabilistic estimates contributed by each contending factors add to the risk. Eventually we have HDN as a forward and backward computed results.

Section B

  1. ‘HEART ATTACK’ | ‘Age(years)’:=’50-59′>=0.0404, 0.2805
  2. ‘HEART ATTACK’ | ‘History of BP’:=’1′ and ‘Taking BP medication’:=’1′ and ‘History of high cholesterol’:=’1′>=0.1171,0.2927
  3. /’HEART ATTACK’ | ? >=0.0860,1
  4. /’History of BP’:=’1′ | ? >=0.3904,1
  5. /’Taking BP medication’:=’1′ | ? >=0.2910,1
  6. /’History of high cholesterol’:=’1′ | ? >=0.2911,1
  7. #    NET = {166.34%,8.21%}
  8. #        = iota(166.34%) + iota*(8.21%)
  9. #    where real part = existential joint probability component = 87.275%
  10. #    where imaginary part = universal purely conditional component = h*79.065%

 WHAT IS THE PROBABILITY OF not having a history of high cholesterol, for a female patient age 50-59 taking diabetes medication and having a BMI of 30-39 and HDL 50-59, GIVEN THAT the systolic BP is 130-139 and Non-HDL is 120-129? Also implies the question with the lower case parts of the first and fourth lines switched.

The above is easily translated into Dirac brakets, manually or automatically, as shown below. To get the probabilities for those brakets, the DiracBuilder module looks for tags outputted from DiracMiner into Semantic Lake that carry the required information. Here it finds four relevant tags. One corresponds to the complex conjugate of the implied corresponding query tag, e.g. we looked for A, B | C> and found C | B, A>, but we merely imagine switching the Pfwd and Pbwd that it presents. The Query as HDN would be as:-

‘History of high cholesterol’:=’0′ |’BMI’:=’30-39′ and ‘HDL(mg/dl)’:=’50-59′> = Pfwd 70.87, Pbwd 6.65

‘BMI’:=’30-39′ | ‘Systolic BP(mmHg)’:=’130-139′>

= Pfwd 46.44, Pbwd 28.51

‘HDL(mg/dl)’:=’50-59′ | ‘Taking diabetes medication’:=’1′>

= Pfwd 25.11, Pbwd 14.92

‘Taking diabetes medication’:=’1′ | ‘Male’:=’0′ and ‘Age(years)’:=’50-59′>

= Pfwd 11.69, Pbwd 22.91

‘Male’:=’0′ and ‘Age(years)’:=’50-59′ | ‘Non-HDL(mg/dl)’:=’120-129′>

= Pfwd 20.81, Pbwd 8.09

We multiply the probabilities in each direction, 70.87 x 46.44 etc. to the left, and 6.65 x 28.51 etc. to the right. The answers are 0.2010%, 0.0052% respectively, i.e. the dual – {Pfwd 0.2010%, Pbwd 0.0052%}.

Statistically caused, and so perhaps then Uncaused and neither Determined nor Pre-Determined

Bioingine_HIMSS16._HDN
Cogito ergo sum [a] is a  Latin philosophical  proposition by  René Descartes usually translated into English as ” I think, therefore I am“.

Cartesianism – is usually understood as deterministic, i.e. about  definite answers, deductive logic . Actually Descartes held that all existence consists in three distinct substances, each with its own essence (https://en.wikipedia.org/wiki/Cartesianism)

  1. matter, possessing extension in three dimensions
  2. mind, possessing self-conscious thought
  3. God, possessing necessary existence

It was arguably a bold attempt to be coldly rational by divorcing the first of these from our notions of the mental and the spiritual, that the first was divorced from the influence of the mental and the spiritual. It is only a convenient partitioning that limits our perceptions.

Descartes explained, “We cannot doubt of our existence while we doubt.” A fuller form, dubito, ergo cogito, ergo sum (“I doubt, therefore I think, therefore I am”)A non-Cartesian view would be that such things are not distinct, any more than matter and energy is distinct. There is only information, and in the sense that it has meaning, impact on us, actionability and consequences,  we mean not a string of bits or the stratum that holds them, but its organization into knowledge and wisdom.)

Insomuch that belief, cause, and  probability are all one, this has much to do with the philosophy of  the Presbyterian minister Thomas Bayes https://en.wikipedia.org/wiki/Thomas_Bayes, which has little to do with the popular Bayes Net, merely a use of conditional probabilities  constrained such that lements of reasoning imply a knowledge network that must be a directed acyclic graph.

In modern physics, many hold the view that all three of Descartes distinct substances are human perceptions within the universe as a giant computation device in which what we perceive as the external world, our minds, and the mind we exist within, are the same.

In philosophy, the new wave non-Cartesian was Edward Jonathan Lowe (1950 – 2014), usually cited as E. J. Lowe but known personally as Jonathan Lowe, was a British philosopher and academic. Oxford-trained, he was Professor of Philosophy at Durham University, England. He developed a version of psychophysical dualism that he called non-Cartesian substance dualism.

“It is an interactionist substance dualism. (Cf. John Eccles and early Karl Popper.)… Lowe argued, however,

That events (both mental and physical) should properly not be thought of as  causes, because only actors (human or animal agents – or inanimate physical agents) can cause things. Events are more properly simply happenings, some caused, some uncaused. [As] quantum indeterminism states, some are only statistically caused, and so  perhaps then uncaused and neither determined nor pre-determined.”(http://www.philosophyonline.co.uk/oldsite/pom/pom_non_cartesian_dualism.htm,https://en.wikipedia.org/wiki/E._J._Lowe_(philosopher).

Our approach based on the mathematics of physics and in particular Paul A. M. Dirac and Albert Einstein’s’ view of space-time as one says

that there is no cause and effect in the specific sense that, in a network of interaction the illusion of non-locality is maintained by the fact that who was the cause of preparing a state and who observed it is not meaningful. Rather, all  co-exists in space-time, so that while   the Bayes Net (again, little to with Bayes) denies  that something can cause something that causes something that causes the  original cause, such statements are meaningless, in that all things interact in space and time, and cyclic effects abound in nature as they do in quantum mechanics,  and in our reasoning about them..

Our approach  additionally includes inductive logic and is combination of number theory, information theory,  quantum mechanics and knowledge / extraction Bayesian statistics that have all to do with processing the interactions between things in a larger knowledge picture.  Only such a view

  • can solve the difficult classes of problems:-
  • creates semantic data lake
  • can do cyclic Bayesian computations
  • can do reasoning and inference to reach a degree of knowledge
  • can manage the non-predicated and probabilistic, and  discover the possible hypotheses existing and new.