BME100 f2016:Group3 W8AM L6

From OpenWetWare
Jump to: navigation, search
Owwnotebook icon.png BME 100 Fall 2016 Home
People
Lab Write-Up 1 | Lab Write-Up 2 | Lab Write-Up 3
Lab Write-Up 4 | Lab Write-Up 5 | Lab Write-Up 6
Course Logistics For Instructors
Photos
Wiki Editing Help
BME494 Asu logo.png


OUR COMPANY

Name: Madison Kleszcz
Name: Kayla Pociejewski
Name: Bianca Silva Juarez
Name:Midori King
Name: Alexandria Castillo

Our Brand Name

LAB 6 WRITE-UP

Bayesian Statistics

Overview of the Original Diagnosis System

The diagnosis system in BME100 was divided into seventeen groups diagnosing two patients each, which provided data with a total of 34 patients. Each group approximately has five to six individuals, allowing each group to delegate different positions and duties to create an efficient system not only within the class, but also within the group to obtain data for the population being studied. In our actual collection of data, however, two of the groups failed to share the data which means only 30 patients' data was used in calculations. In attempt to prevent error during the experiments, there were three trials for each patient, and a positive and a negative control to use as our gold standard during calculations. This along with the technologies used in ImageJ contributed to the overall results of the experiment, however inconsistencies still occurred. According to the class's final results, only 69% of patients were correctly diagnosed with the disease regarding a positive PCR result, while only 63% of patients were correctly identified as not having the disease regarding a negative PCR result. As stated before, two groups' data was not included in the shared data, shrinking the total population by four patients. In addition, there were six total inconclusive results that could have greatly shifted the overall results. Inconclusive results could have been due to problems in the overall design of the PCR machine, or also the process by which the group completed the experiment with the fluorimeter. Some of the challenges that could have affected the data may have included using contaminated tips during the PCR tube creation processes; using the wrong settings when calibrating the PCR machine; having different distances between the drops and the camera in the fluorimeter; not having the appropriate camera settings to correctly capture the drops' identities; allowing too much light to come in when taking each picture of the drops; and confusing which drop was which when transferring the images onto the computer to complete the ImageJ data. Overall, the lack of experience in handling the PCR machine and the fluorometer most likely led to a number of errors in the data. While my group received accurate data in comparison to the doctor's diagnosis of our patients, we found that most of our difficulties came with the fluorimeter itself. The process of completing the experiment was not the problem, but instead the actual design of the fluorimeter caused us the most difficulty. Making sure that the slide with the drops were a consistent distance each time, making sure that the camera took a clear picture of the drop, making sure that the drop did not move across the slide, and making sure no light came into the fluorimeter were the most flawed parts of our experiment. If the fluorimeter could be better designed, we believe that making a correct diagnosis could be made a lot easier.

What Bayes Statistics Imply about This Diagnostic Approach


The ability of the PCR replicates to correctly describe the sensitivity of the system regarding correctly detecting the SNP disease was relatively high with the results indicating that about 90% of the patients will get a positive test results given that the PCR replicates indicated a positive reaction. The calculation regarding the specificity of the process to detect the disease was nearly 100% . This means that the experimental process is very accurate in detecting whether or not a sample receives a negative test conclusion given a negative diagnostic signal.

The sensitivity of the system to actually detect the disease, however, was not as high at about 65% accuracy. Likewise, the ability to correctly detect if a patient will not develop the disease with a negative PCR result was only about 67%. While both of these results are similar, the ultimate result would be improved if the error margin was decreased for a false negative because this could lead to delayed treatment for the patient. A false positive is inconvenient for the patient, however, it is better to think you have a disease and frequently go to the doctor for more tests than thinking you are healthy and not receiving any treatment whatsoever.

One possible source of human error that could have greatly affected the Bayes values in a negative way could be using contaminated micropipette tips to create the PCR samples. This mistake would have caused inconclusive results that could have skewed whether or not a projection was positive or negative for the disease. Another source of error could be the limitations of the phone's camera that we used to capture the pictures of the drops. While certain settings were required to correctly capture the identity of each drop, the camera that we used did not have all of the ideal settings to ensure proper results. In addition, the camera had difficulty focusing on the drop since the distance between was so small which could have also made our results in ImageJ not as accurate. Lastly, another error that could have negatively affect the Bayes values could have been the design of the fluorimeter in which complete darkness was not possible when a hand had to placed inside to take the picture. This conflict could have easily made the coloring of the drops less intense and thus could have been mistaken for a different conclusion in the ImageJ program.

Intro to Computer-Aided Design

3D Modeling
During this experiment, our team used Tinkercad. Tinkercad is a web based design for students who are new to designing 3D models online without any previous design experience. It is easy to use and user friendly. It has a number of tools that allow you to work with the program efficiently making the process easier to learn. The simplicity of the program made it easy to use and work with without being confused. Designing a new fluorimeter with a clear solid model was important to our design process because we wanted to show our new feature that was added to the original design. The camera feature is placed on the inside of the fluorimeter and without the tools available in Tinkered, our design would be unclear in our rough sketch. The clear base and solid structure allows for our design to be viewed.


Our Design

2016-11-09.png


Our design consists of an addition of a camera built within the fluorimeter itself. This design was chosen based on the act that we had issues trying to use a cellular device to take photos. As a group we decided that this was an issue based upon the fact that a cell phone is not set up properly to take the images needed for the experiment. It was difficult to set apart what photos were for which data point based upon the fact that the pictures looked very similar to one another. A built in camera would decrease the amount of space for a physical error within the experiment. With the addition of a camera, the design would be compatible with an app that allows the camera to be controlled via bluetooth. The app would control when he pictures would be taken and would allow you to name the picture files. This prohibits there to be more room for error since the images are labeled after the picture is captured, which allows you to work with the solutions and the data collected in a more efficient manner. The time would be reduced because there would be no confusion as to which image captured belongs with the different solutions tested.


Feature 1: Consumables

Our main focus is on redesigning the fluorimeter. Our new design consists of an addition of a built in camera and a memory card that stores the pictures that are taken. Since the camera is a new design aspect, you would think that the packaging would change dramatically but in our case, the camera will be built in to the fluorimeter itself. This allows the packaging to stay consistent with the previous packaging. Moreover, all the same consumables for the PCR machine process remains the same including the PCR tubes, the micropipette tips, the PCR mix, primer solution, SYBR Green solution, and the buffer. In addition, the glass slides used to hold the drops on the fluorimeter are still needed to receive accurate results for each different reagent.

Feature 2: Hardware - PCR Machine & Fluorimeter

The fluorimeter design is controlled similarly to the previous design. The main difference is that the camera is controlled by a bluetooth app that allows the user to correctly label each picture. In addition, our fluorimeter includes a built in camera at a specific distance from the device so that each image is perfectly consistent with one another. This feature also allows the picture of the drop to be captured in complete darkness to ensure the best results possible. The Open PCR machine works the same way as the original setup. In the original procedure, the OpenPCR program was used to run a heating and cooling program on the thermal cycler. The heated lid was set at 100 degrees celsius, the initial step was set at 95 degrees celsius for 2 min, the number of cycles at 25 with a denature at 95 degrees celsius for 30 seconds, anneal at fifty-seven degrees celsius for 30 seconds and extend at seventy-two degrees celsius for 30 seconds. The final step for the OpenPRC program was to run for 2 minutes at seventy-two degrees celsius and a final hold at four degrees celsius. This same program was run with our new design.


Our group determined that the major weakness in the fluorimeter was the failure of a phone's camera to take consistent photos. The addition of a camera would resolve any possible errors that could take place when trying to capture a photo of each data point. In addition, the organization of the images also was difficult to keep consistent. Possible error could have come from mistaking one image for another because there was no way to label the pictures on a cellular device. Thus, our new fluorimeter contains a slot for a memory card and a bluetooth feature that connects to an app. This app would allow the user to name or label each individual picture that is being taken. The labeled picture will then be stored on the flash drive and thus no confusion can be made when using ImageJ to calculate the results.