How best PAIRS can be useful?

April 15, 2008

To answer this we need to define what we want from a Diagnostic Decision Support System (DDSS).

(1). We want DDSS to help us in solving difficult cases. If a patient data suggests a couple of diagnoses which can be tested straightforward then we should not waste our time with a DDSS. However, if we suspect that the case is not as straightforward as it looks to be then we definitely need a DDSS. How can one be sure that a case is straightforward or not? Probably by questioning oneself further and seeking answers. One cannot be satisfied totally about the finality of one’s diagnosis since there are several gaps exist in our knowledge which need to be addressed before we arrive at such a conclusion. For example, what are the geographical, gender and age parameters that may influence our diagnosis? Our need of updating such epidemiological data is constrained by one’s likes or dislikes. What if we don’t like such data to influence our diagnostic decision? Can we prove that such a decision is infallible? Well, why not ask a DDSS to help us?

(2). When we arrive at a probable diagnosis of a difficult case, we need to tally it with pathogenesis (infective, neoplastic, degenerative, metabolic or congenital), precipiating cause (obesity, smoking, drugs, diet, alcohol, triggers) and duration (acute, chronic or acute on chronic). Is it systemic or non systemic in its manifestations? Is it practically possible to work on such a data consistently? I think may be not unless we focus on a few disease possibilities. In such a case we definitely need a DDSS.

(3). When we finally zero in onto a diagnosis as a best bet, we need to find an easy way of checking. Easiest way would be to look for some symptoms or signs in our patient that we have missed earlier on. Alternatively, we need to go in for a diagnostic procedure that one can perform to conclusively prove our diagnosis. Either way we need to analyze lot of data that may not be readily available for the doctor. For example, what extra features if we add to the patient data, our diagnosis can effectively change? Such analytical skills are rare and difficult to acquire in medical profession. In such a situation DDSS can help us.

PAIRS is developed to help doctors in the way we described here. So hopefully, one can benefit from using it. In conclusion, we do not claim that PAIRS gives a correct diagnosis for a given patient data. This is not our aim. Our aim is to help doctors in arriving at a correct diagnosis using a tool that can serve the needs of doctors in diagnostic decision process.

Advertisements

SNOMED CT concepts are used for PAIRS

March 28, 2008

A paper on “Using SNOMED CT concepts for PAIRS” by AM Mohan Rao (www.lmspairs.com) is accepted for upcoming KR-MED2008 conference at Tucson, Arizona, 31May 2008.

http://www.kr-med.org/2008/program.html

It describes how PAIRS Natural Language Processor works.

Standardized Nomenclature of Medicine Clinical Terms (SNOMED CT) is developed by College of American Pathologists (CAP) in collaboration with National Health Service (NHS) of UK for last 30 years. Now it is owned by International Health Standards Terminology Development Organization, a consortium of developed countries including USA, UK, Canada, Netherlands, Australia, Denmark and NewZealand.

Logic Medical Systems was granted a license to use SNOMED CT by National Library of Medicine during September 2007.

Use of SNOMED CT as part of NLP for PAIRS enables it gain user confidence.

PAIRS user instructions

March 11, 2008

You can access PAIRS at www.lmspairs.com. Front page centre region shows PAIRS and PAIRS LM image links. Click PAIRS if you want to try diagnostic domain. Click PAIRS LM if you want to see any disease links to abstracts in National Librart of Medicine (NLM) PUBMED site.

PAIRS diagnostic domain asks for a user ID and password. If you are visiting PAIRS first time it is advised to register by clicking new user registration. Free introductory trial for a week is available for new users. Your password will be activated with in 24 hrs.

After entering the userId and password click Login button. You will reach Main page which has several buttons to navigate to. “Single case diagnosis” and “Select case diagnosis” buttons navigate to diagnostic pages where as “Feature-to Feature links” and “Disease-Feature links” lead to database pages of PAIRS.

Click “Single case diagnosis” button and click CPC cases link to download a text file to your computer.

 Open it to select a patient data and copy paste it to the “Enter patient data” text area on the left side of the page.

 Select age, gender and nation and click submit button. In a few seconds your “Final diagnosis” page opens up. You can see the age, gender and geograaphical based diagnosis and age and gender based diagnosis on top.

You can select your options in drop down list for precipitating case, pathogenesis, system and duration appropriate for the patient and select a disease possibility for diagnosis and click submit button.

 You can see the graded possibility for the selected disease appearing below the options list along with procedures to be done to diagnose that particular disease along with links to abstracts in NLM.

 You can change the options to get the data for a different disease. On left side “Look for features” text area gives features to look for in that particular patient for given the diagnosis.

Navigate to  “Select case diagnosis” button. This facility enables one to load all cases of a hospital and analyze them.

As an example, we work with CPC cases you have already downloaded.

In “Multiple Patient Data Entry” screen click Browse button and navigate to the folder where your CPCcases.txt is located.

Click “Attach” button.

You will see the cases listed in the file appearing in the text area. Click “Submit” button.

You will see a number representing total number of cases appearing in the “Number of cases uploaded” region.

Select a case, age, gender and geography as explained for a standalone case and click submit button.

You will see screen navigate to the Final Diagnosis page.

You can see the age, gender and geograaphical based diagnosis and age and gender based diagnosis on top.

You can select your options in drop down list for precipitating case, pathogenesis, system and duration appropriate for the patient and select a disease possibility for diagnosis and click submit button.

 You can see the graded possibility for the selected disease appearing below the options list along with procedures to be done to diagnose that particular disease along with links to abstracts in NLM.

You can change the options to get the data for a different disease.Look for features in text area on left side gives features to look for in the patient for that given diagnosis.

Navigate to “Disease-Feature links” and enter “ASD” in left side text field and click “Go” button.

Select “Atrial Septal Defect” that appears in a text area below and click submit button.

You can see disease-feature links appearing in a table below.

Type “Tachycardia” in text field on right side and click “Go button”. Select “Supra ventricular tachycardia” in the text area below and click submit button. You can see feature-disease links appearing in a table below.

Navigate to “Feature-feature links” and select “Acne” in text area on left side and click “Go” button.

Several features fill text area on the right side. Select “anemia” and click submit button. You will see several diseases that have both “acne” and “anemia” as features fill a table.

You can log out of PAIRS by clicking on logout button on right upper side.

Click PAIRS-LM page and click on “Access common diseases” button. You will see many diseases appearing in the page that opens. select any one of them and click on links of your interest to navigate to an abstract in PUBMED site. Similarly you can view data for rare diseases by selecting “Access rare diseases” button in PAIRS-LM page.

PAIRS (Physician Assistant Artificial Intelligence Reference System)

March 8, 2008

Welcome to PAIRS at www.lmspairs.com a great site for instant medical info. It is a great web site for doctors, patients and any one seeking a medical opinion or information. Ofcourse, PAIRS does not offer to give a medical diagnosis but can help in suggesting a diagnosis using right technology. You can enter a patient data in a text box and click. PAIRS gives you diagnosis based on an Artificial Intelligence method.