posted 09-21-2010 09:02 PM            

Be gentle... if it sucks I don’t want to be treated like a piñata. I think we did a pretty good job emulating the arm cuff however I’m not an examiner and it’s your opinion that I would like to hear.

posted 09-22-2010 07:10 AM            

My mind is limited to innovation and building things. I’m also not a big advocate for in-house research. Research and validation is for greater minds. I trust this to professionals that ideally do not have a monetary interest in achieving only positive results. I’m always open to co-operation if anyone is interested in doing validation studies. If they need Limestone’s assistance and resources please call us and let’s talk.

posted 09-22-2010 08:45 AM            

Continued observation and evaluation of both components, run simultaneously would tell one whether or not the "new" component performs as well or better than the "old" component. In reality, the recorded data is the same just obtained using a different location and/or device.

Many of us switched without hesitation from EDA finger plates to stick on electrodes and from analog mechanical components to computerized electronic components without the need for validity research.

I would suggest that one run both and observe both to determine if the finger cuff provides the same visible reaction tracing to applied stimulus as the conventional arm cuff.

If it does, and we are not losing data, then the finger cuff is certainly less intrusive and would be a better choice.

I'm not in the camp that believes pressure in the arm cuff is a good thing. If we are looking for evidence of differential salience, introducing discomfort as a variable would not seem to be desirable. Any time we can remove a non-stimulus induced variable, it seems it would result in less signal noise and therefore more viable data.

For the first 7 years of my polygraph career, I ran an analog instrument with both a mechanical and an electronic cardio component attached to the same cuff. We were required to do so because it was believed that there was criteria displayed from the mechanical component that was not being faithfully displayed in the electronic component. When battery operated, portable instruments became popular, we simply dropped the mechanical component to make room for the battery pack without a second thought.

posted 09-22-2010 07:59 PM            

quote:

---

Many of us switched without hesitation from EDA finger plates to stick on electrodes and from analog mechanical components to computerized electronic components without the need for validity research.

---

The only problem with this line of reasoning is that both electrodes measure EDA, and the computerized instrument had been used in the lab before we got it, so we knew both worked. (EDA had been recorded in all kinds of ways. Most of the researchers used to make their own gel and components, so it's not like what we got was really new and untested.)

The question is whether or not the thumb cuff actually measures what the cardio cuff measures. Somebody who was collecting the data with both told me that, for the most part, the two were the same, but sometimes the arm cuff would show an increase while the other, a decrease, and vice versa (which, interestingly, is what you tend to see in the plethysmograph if that waveform isn't filtered out).

posted 09-23-2010 09:45 AM            

quote:

---

I've run cardio cuffs on the upper arm, forearm, wrist and mid-calf. The tracings may be different but the different is in both the comparisons and the relevants. I don't see how a thumb cuff would be any different.

---

posted 09-23-2010 11:04 AM            

Skip Webb appears correct in this discussion. Bottom line is the thumb cuff can probably work just as well and provides similar information as does the arm-cuff if we design a good one. Before he died, Dr. Geddes of Purdue University, and I discussed this issue. He believed the measurement at the thumb was the same relative blood pressure change we can measure at the upper arm or at the forearm or at the wrist. The current designs are plagued by movement and temperature related artifacts. There is so little air in the bladder that small volume changes that do not affect the big cuff, have a dramatic effect on the smaller bladder cuff. Anyone with one can test this by inflating the cuff and breathing hot air in to it and watch the tracing rise. Another problem is the sensitivity of the cuff. The primary arteries of the thumb produce a considerably smaller pressure wave than say the brachial artery. That is why you have to click that setting in Lafayette software to essentially amplify the signal. Amplifying the signal helps but it is a double-edge sword. It also amplifies distortions. Also, the cuff material is thick and does not allow the transducer to adequately detect the pressure wave. Hope this helps.

Bootstrap Correlation Study of Computer Algorithm Scores for
Finger Cuff and Arm Cuff Data Using Directed Lie Screening Polygraph Exams
and the Objective Scoring System version 3

xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx

Raymond Nelson, Mark Handler, Donald Krapohl and Sheila Thomas

posted 09-23-2010 08:37 PM            

Firstly, we need to know if the Lafayette and Limestone finger cuffs are the same.

Secondly, we have to consider whether or not the internal components of the data acquisition systems are the same.

Thirdly, the mm Hg in a cuff can change the data received.

Lastly, we need to consider the subject the measurement came from. For instance, a subject with hypertension, type 2 diabetes, or even reduced blood flow to the arteries of the finger due to sympathetic responses.

posted 09-24-2010 05:19 AM            

"There are many things to consider here, and these are but a small list.

Firstly, we need to know if the Lafayette and Limestone finger cuffs are the same."

No the cuffs are not the same but they operate on the same principle, and so design construction can matter. We were given data collected with the Limestone cuff but did not analyze it.

"Secondly, we have to consider whether or not the internal components of the data acquisition systems are the same."

Once again, same principle of measurement. I don't see why internal components of the DAS unit should matter. As long as they transduce the signal accurately, who cares?

"Thirdly, the mm Hg in a cuff can change the data received."

This is a very confusing statement so I will ask you to clarify. Pressure in the cuff will determine how close pressure is to MAP (mean arterial blood pressure) which has been found to be approximately maximum oscillation pressure. If the pressure in the cuff is above MAP then the amplitude of the tracing decreases with an increase in pressure as the cuff pressure and MAP pressure approach one another. If the converse is true, the amplitude will increase. This has little effect on what is received so long as transducer sensitivity is sufficient to detect the pressure wave under the cuff, and most instruments are.

Please see the paper Dr. Geddes, Joel Reicherter and I wrote on this subject.

Handler, M. Geddes, L.A., & Reicherter, J. (2007). A Discussion of Two Diagnostic Features of the Polygraph Cardiovascular Channel. Polygraph, 36, (2), 70-85.

Lastly, we need to consider the subject the measurement came from. For instance, a subject with hypertension, type 2 diabetes, or even reduced blood flow to the arteries of the finger due to sympathetic responses.

posted 09-24-2010 06:30 AM            

Thanks.

posted 09-24-2010 09:06 AM            

The intention of my last post was to note variables for future address.

The sensitivity of both the instrumentation and the component can have an effect on the data received. What input they are placed on the DAS could also have an effect on the data also. Whether these things will significantly affect the outcome of the observation or not should be investigated when possible, not assumed.

The mm Hg of the cuff, as you investigated and reported about, does have an effect on the observed data. Therefore, finger and arm cuffs that are not equally adjusted may produce diverging or converging tracings.

My point regarding the medical issues was not that they would produce inadequate tracings, but that finger arterial blood flow is affected by certain medical conditions and under stress or physical exertion can cause differences in the data received from the arteries in the arm and finger. These differences could affect the outcome and, thus, is a variable that should be addressed.

Subjects under stress are expected to have reduced blood flow to the extremities. One can expect that differences in data could be more noticeable depending on the proximity of the source of collection to the heart. Again, this is just another variable for us to think about for future research.

I am elated that you and others are addressing this and commend you.

We have begun to look at this and other issues in our field as scientists. Insomuch as we are, we need to look at both the research that has been conducted and that which may be conducted in the future and address those areas that could potentially effect the observations. If we can control for these things and are able to address them with further observation, it is our responsibility to suggest the potential issues that should be addressed in future research.

posted 09-24-2010 11:15 AM            

Conduct a sufficient number of field examinations by different examiners with diverse experience levels which by design would be under the same environmental and biological conditions for the test location and subject.

Have OSS3 evaluate the charts once using one component tracing and then using only the other component tracing.

Examinee medical conditions or testing environment would not appear to have a differential influence under this type of comparison.

Either the finger cuff, properly applied, produces the same result or not. This type of testing wouldn't tell us if the finger cuff is superior to the upper arm cuff but it would answer the primary issue of "Is the finger cuff equal to the arm cuff as a reaction capture device for the data we evaluate?"

The question then becomes "In the absence of the traditional upper arm cuff does the data collected change?" and if so "Does the change in data adversely effect the identification capability of the polygraph test.

Some have argued that cuff pressure and the resulting discomfort resulting from it has an effect on the polygraph test and the level or degree of reaction we observe.

posted 09-24-2010 05:00 PM            

"Either the finger cuff, properly applied, produces the same result or not. This type of testing wouldn't tell us if the finger cuff is superior to the upper arm cuff but it would answer the primary issue of "Is the finger cuff equal to the arm cuff as a reaction capture device for the data we evaluate?"

The question then becomes "In the absence of the traditional upper arm cuff does the data collected change?" and if so "Does the change in data adversely effect the identification capability of the polygraph test."

Data from the finger cuff, if effectively captured, either supplies redundant or new information. That information either produces the same, better or worse results. That is the first question to be answered as long as the data are usable.

posted 09-25-2010 11:52 AM            

I too don't mind putting a little non-harmful pressure on the examinee's arm. The arm cuff does not, as we use it, completely occlude circulation, and no-body's are has ever actually fallen off during or as a result of a polygraph test. Despite all the drama from the examinees, there is no pile of arms in the back-rooms of the polygraph testing suites.

We have data on both available finger cuffs, and the neo-natal cuff, used on the thumb, actually appeared to perform the best.

r = .69 is not bad for the test data, and and r = .95 for the test result is excellent.

The problem is that finger cuff data is fragile and is very easily disrupted. When it works it works just fine. If you read the abstract you'll see that about ~25% of the data were unusable, compared to about ~12% for the arm cuff. This is like buying new tires for your car, but only 3 of them will work. I think most of us would like to see better reliability than this - and this is with the best of the available cuffs.

Another problem is normative data, and the absence of statistical norms for the thumb cuff. We could re-norm the algorithm and our manual scoring methods (most of which actually have no normative data anyway), but it would be better to wait until we have a more robust solution or we'll just have to repeat all of our efforts in the near future.

Before we consider a new component to be a direct replacement, it would be nice to see a component correlation of .9 or so. Present correlations may improve with a non-parametric transformation method. OSS-3 is a parametric model, but our manual scores for ESS are non-parametric. 7 position manual scores are not non-parametric, because they make the assumption that the data have a linear parametric shape. However, 7 position scores are not actually parametric because physiological response data is non-linear. So 7-position scores are neither parametric nor non-parametric - they are negligent of our measurement and statisical principles. So, we have more questions to answer.

When we see robust operation that meets or exceeds what the arm cuff does, and when the sensor correlation exceeds .9, then we might begin to consider an alternative cuff to be a direct replacement or reasonable substitution to the arm cuff.

In the meantime, the traditional arm cuff appears to be the most robust sensor, and we have the largest knowledge base underneath it.

So, consider this:

How can we justify - while goofing around with people's lives, community safety, and national security - doing something that we know is not the best?

Gimme and arm-cuff please.

as always,

.02

r

------------------
"Gentlemen, you can't fight in here. This is the war room."
--(Stanley Kubrick/Peter Sellers - Dr. Strangelove, 1964)

posted 09-25-2010 07:23 PM            

I can tell you it does not correlate like we see in the advertising pictures. I can speak from experience. I had to go through ALL of the exam charts and assess the, artifact them and assign a subjective "grade" to them. Then I had to strip the raw data from the exams and enter it into spreadsheets. I worked a week straight even with the generous help of the Lafayette software engineers to get the data between X and XX set up to analyze. For me the finger cuff waveform display is harder to evaluate. It seems less responsive and does not always follow the arm cuff wave form display.