IT'S BEGINNING TO LOOK A LOT LIKE CHRISTMAS (BME edition)
Daniel Beltran
Matthew Castile
Catalina Pardo
Brenna Toshner
Nivina Warner
Our Brand Name
LAB 6 WRITE-UP
Bayesian Statistics
Overview of the Original Diagnosis System
In this lab, BME students tested patients with a disease-specific primer through the process known as polymerase chain reaction (PCR) to see if the patients had the disease associated with the SNP. 16 teams of about 3-5 members each were given samples of two different patients and a positive and negative control. The positive control was a sample obtained from a patient that did have the disease while the negative was from a patient that was disease-free; the results obtained from the PCR lab were to be compared with these samples. In order to prevent error or, in this case, a misdiagnosis, there were three replicates for the sample of the patient that were to be tested and examined. After the PCR was completed, the students used a fluorimeter and ImageJ to examine these samples. Before analyzing the samples obtained from PCR, calibration samples were used and analyzed to ensure the accuracy of the fluorimeter, phone camera, and ImageJ. Afterwards, the same amount of each sample was mixed with SYBR Green and placed in a new part of the slide, so as to stay consistent and prevent contamination. Several pictures were to be taken of the drops from the samples and analyzed with ImageJ. The phone that was used to take the pictures of the drops was always placed at about the same distance and height from the sample. Three pictures were taken of each of the drops to decrease the error margin. Since there were a total of 16 teams, each with two patients and three different samples for each, there were a total of 96 PCRs. After analysis of the class' samples, it was concluded that 56 patients were tested positive and 39 were tested negative. As for each of the team's conclusions for each patient, 17 concluded that the patient was positive, 13 were tested negative, and 2 were inconclusive. When compared to the actual diagnosis for each patient, 15 were actually positive and 17 were negative meaning that there was about an 18.75% error in our conclusions. Our data may have been affected during the fluorimeter process as we started with a PCR sample before performing the calibration. Additionally, a drop of a PCR sample fell off the slide and so it may have contaminated the slide or another sample.
What Bayes Statistics Imply about This Diagnostic Approach Calculation 1 tells us the probability of a patient that tested positive in the PCR would also be tested positive in the final test conclusion. This probability, about 80%, is quite reliable, but can possibly be better. Calculation 2 tells us the probability that a patient that obtained a negative diagnostic signal will also test negative in the final conclusion. This percentage, over 90%, is far more reliable than the one from other calculations, but still not as close to 100% accurate as it could be.
Calculation 3 gives us the probability that a patient that tested positive in the final will develop the disease. This percentage, roughly higher than 50%, is not very reliable. Unlike the other calculations, the results from calculation 4 tell us that there is a 100% chance that a patient which tested negative in the final test conclusion will not develop the disease. This calculation, in other words, tells us that this conclusion is 100% reliable.
Some possible sources of error that may have hindered the Bayes values are; light exposure in the fluorimeter, contamination during the prep in the slides, and different amounts of the sample/SYBR Green 1 solution when pipetting. The light inside the fluorimeter may have messed up the pictures that were taken, which in turn, gave us faulty results from ImageJ. When adding the samples on the slides for examination, contamination may have given a false positive or false negative later on. Finally, different amounts of the solutions when pipetting and mixing the samples may have given erroneous concentrations that were analyzed.
Intro to Computer-Aided Design
3D Modeling
For this lab, our team decided to use TinkerCAD due to the fact that the majority of our team uses Mac computers. TinkerCAD turned out to be perfect for this lab as the pre-made designs helped us to create the shape of the new fluorimeter model more efficiently. Most of the shapes we planned on using were simple, as well as TinkerCAD's vibrant colors allowed for a deeper contrast of what we decided to bring out in our changes. The controls were intuitive, in fact, we didn't go through the tutorial at all. The buttons were large and fairly easy to recognize what their general function was and after 10 minutes of clicking around we had already figured out every button we needed to create a simple design of what we plan on implementing. TinkerCAD does have a tutorial that, after about 5 minutes of not being able to shift everything into view using the mouse, was very helpful in getting the right shots for our design. For what we did with this design (simple shape, simple concept, no gears, etc), it was perfect, however, for more complicated designs, we don't believe TinkerCAD would do the job. Overall, we would rate the experience of using TinkerCAD as an 8.5/10.
Our Design
We decided to improve the Fluorimeter Machine. During our time doing this PCR lab we found it hard to balance our iPhone 7 plus, the box let in too much light, each picture had a different focus on it, and changing out the slides each time proved to be a challenge for keeping everything consistent. In our design, we eliminated the need for the group to struggle with setting up their phone. On the side of the box is an adjustable phone holder with a lens implanted on the side of the box. Since the holder is adjustable, it does not matter how big or small the phone is, as well as which side the camera is on because the holder would adjust to fit the groups' needs. The lens would not shift, it would be perfectly aligned with the drop the group is testing. The lens would eliminate the need for changing ISO, shutter speed, etc. This would allow consistency amongst all pictures, as well as a more intuitive experience for the group. This would save time and allow for more efficiency when tasked with taking over 20 different pictures in the lab time allotted, it would also help when analyzing pictures during post-lab. We also decided to include a long, moveable 3D-printed plastic slide that does the job of holding the PCR experiment. The group simply Micro-Pipets onto the slide like a glass slide, except with our new design the group must pipet at least 3 of the samples they would like to look at first. Then, they slide the long plastic through the rectangle big enough to fit the slide and solution drops. The rectangle is perfectly aligned with the base that holds the light. When the solution being photographed makes it to the light, the group snaps the photos, puts the next solution on the slide (outside of the box), then slides to the next solution in line, already on the slide, and repeats the process. This eliminates the need for opening and closing the top of the Fluorimeter Machine and possibly ruining the controls already set up. We think these adjustments will allow for much greater efficiency and understanding on the students' side.
Feature 1: Consumables
Consumable Product Kit List:
PCR Mix
Primer Solutions
SYBR Green Solution
Buffer
Plastic Tubes
Pipette
Pipette Tips
3D printed plastic slides (company made)
The material of the slides was changed from glass to 3D-Printed Plastic. This change allows for the reuse of slides, since the 3D-Printed Plastic is washable, tough and reusable. The length of the newly designed slides are also longer, which allows the experimenter to keep the fluorimeter box closed for the duration of their experiment. The original design of the slides did not allow for reuse and the experimenter would often have to switch out slides during the same experiment which allowed for possible errors in the lab.
Feature 2: Hardware - Fluorimeter
The way we redesigned the Fluorimeter System, was by including a camera lens that will be compatible with any type of smartphone. This was done in order to reduce the amount of outside light and to ensure that each picture will be of the same quality definition. The system will come with an opening that the camera lens will be placed in, which will make the process easier. There will also be a set height for where the drops will be placed so they will always be the exact same distance from the camera. Since it is so practicable, schools in need of an easy-to-use, quality system will be the main target. It is also more compact than a regular fluorimeter which might encourage consumers to buy more since they take up less space than a regular one. Also, because of it being exclusively an online product, it will ensure availability.
BME 100 Fall 2017
