3.0 OnICS/iNOTZ solution

 

The OnICS solution for electronic records outlined in section 2, and developed in our iNOTZ program, was inspired by the work of the Bletchley Park code breakers and the DNA pioneers. Welshman, at Bletchley Park, developed traffic analysis techniques in 1940 just before Turing cracked the Enigma code. Welshman’s work was first published in 1982 but was banned in the USA under secrecy laws. However, it was recently published in 2018 in “The Hut Six story” M & M Baldwin ISBN 978-0-947712-34-1. Turing’s paper (1937) “On computable numbers with an application to the Entsceidumsproblem” laid the basis for deciphering the Enigma code (Proc. London Mathematical Soc. Ser. 2, 42.230. Paper read to the Society 12 November 1936). A subsequent paper (Turing, AM “Computing Machinery and Intelligence” Mind 59, 433-460, 1950) gave birth to the discipline of artificial intelligence (AI). His two papers established Turing as the father of modern computing and the second Turing paper (1950) contains a section entitled “The Imitation Game”. This, together with the structure of DNA (Watson, JD and Crick FH “Molecular structure of nucleic acids; a structure for deoxyribonucleic acid.” Nature 171, 737-738, 1953) forms the basis for our Digital Linguistics record system using single bit logic.

3.1) Definition of data points (in Oncology)

These are single data items that can be a word or phrase.

3.1.1) A symptom, (lump, bleeding, cough, etc.)

3.1.2) The organ involved and subsite

3.1.3) Symptom duration

3.1.4) Every examination finding (size, mobility, ulceration, etc.)

3.1.5) Investigation requests and results

In dictated notes the data-points are embedded in the morass of dictation and consequently they are ‘lost’ or, at best, difficult to find and extract. Such notes are always incomplete and never fully structured.

3.2) Digital Linguistics 

Covered by US patent US 10,529, 452 B2 : submitted May 2011 granted November 2019

The possibilities available for electronic annotation systems (section 1.1) include conventional dictation, voice recognition, templates & neural nets. The last of these is attractive but, we have to ask the question:

Why use a computer neural net that needs training?

And, instead of training a neural net de novo we use the output from neural nets that have already been well trained. These are groups of specialist Oncology Consultants who, between them, have over 90 years clinical experience in the UK and US.

3.2.1) DEFINITION: Digital Linguistics is the generation of fluent text from descriptive words and/or phrases selected from menus using implied grammar to impart meaning. Grammar is derived from the context and sequence within which the descriptive words and/or phrases are presented.

3.2.2) Digital data point input – the menu system

3.2.2.1) Abandon the concept of “fields” for dictated clinical input

3.2.2.2) Generate menus containing descriptive words and phrases in a sequence the physician would be expecting, in order to imitate the clinical process (“The Imitation Game”, in Turing, 1950).

As an example, five (5) menus are required for the following sentence.

“The patient presented with a lump in the upper outer quadrant of the left breast

The 1st menu does not contain an item that appears in the sentence but, this is designed to cut down on the number of items in the 2nd menu. If we had started without the region menu, we would have needed 70+ items, one for each organ, in the next menu. Having selected “chest” from the region menu the 2nd menu contains “breast” and the four other items. Menus 3 and 4 are engaged when “breast” is selected as they contain the breast symptoms and the subsite in the breast. The last menu contains the side of the lesion.

 

 

 

 

3.2.2.3) Each menu item is a data-point, defined as a primary data unit (PDU), that is represented by a single bit.

3.2.3) Generating text from the selected menu items

In generating text, we have to appreciate that the digital data must be collected in an order that is necessary and “convenient” for computer input (see above). However, the sequence for collecting the data is NOT always the same as the sequence that has to be used for generating the text. This is shown in the following schematic:

If we generate the sentence in the order of data collection we get:-

“The patient presented with a breast lump in the upper outer quadrant on the left.”

This makes linguistic sense BUT it is cumbersome AND ambiguous, because does left refer to the breast or the upper outer quadrant that is on the left of the left breast.

3.2.4) Single bit logic - the Digital Linguistic solution

3.2.4.1) Keep the data-points in the menus and NOT in the data base

3.2.4.2) Item selection switches a bit in a given word from 0 to 1

3.2.4.3) Each bit set to 1 now represents the address to its menu item

3.2.4.4) Each 0 represents the address to its unselected item hence, “negative” data are recorded in parallel with the “positive” data

3.2.4.5) Each computer word can hold up to 32 units of data, PDUs, as opposed to just four characters with conventional systems

3.2.5) Data Storage – HIPPA compliant

3.2.5.1) Text generated by our Digital Linguistics system is inserted into a conventional EMR data base as a PDF text file by the iNOTZ program.

3.2.5.2) Digital Data are recorded in parallel, but separate from, the text (see appendix 7.3).

3.2.6) Identification, Data Tracking and Work-Flow Analysis

Image recognition technology within iNOTZ can be used for patient and/or physician/user ID and also for photography of superficial lesions and radiation treatment markings. Data tracking by message frequency analysis is recorded by methods similar to those used by Welshman (1982, 2018). Bar coding introduced by Silver, B. and Woodland, N.J. (“Classifying Apparatus and Method” US patent 2,612,994, October 1949) and/or screen signature are available for physician ID that can be used for paperless operations.

For example: - Every time you purchase a bar-coded item with your credit card some computer system somewhere “knows” who you are, what you bought, where and when you bought it and for how much.

3.3) System Schematic

A schematic of the OnICS iNOTZ Local Area Network within a practice is shown as follows: