Biostatistics Shared Resource

Transcript

• 00:01OK, good afternoon, everyone my name is Melinda Irwin. I'm associate director for population Sciences in the Yale Cancer Center. So I'm fortunate to introduce our two speakers today. But before we do so. I just wanted to remind you. I didn't know if innovation summit a spectacular who went to this last year raise your hand.
• 00:30OK, well, you gotta go this year. It's really spectacular that coming up in about a week and a half but you can submit a poster application and here is the information it's do.
• 00:43Battle.
• 00:44OK, like that says April 17.
• 00:48OK.
• 00:51OK, any other comments you want to make.
• 00:561000.
• 00:58It.yeahitreallyisaphenomenonsosubmitted@yellinnovationsummit.com OK great OK, so we have 2 talks. Today, the 1st is Steven MA is a professor of biostatistics in the Yale School of Public Health. He also directs the Yale Cancer Center Biostatistics in Bioinformatics shared resource, which if you're not aware of that shared resource as well as the other shared resources make sure you go to the L Cancer Center website and read about them.
• 01:29Steven was trained in statistics at University of Wisconsin at Madison and has been here for about 15 years and has had multiple NCI studies funded looking at various aspects involved in various number cancers. I'm not sure which cancers are going to discuss today. But a number of different cancers as well. Some cancer disparities research as well. Thank you.
• 02:08Many thanks for the introduction and also is great pleasure being here. I think I give a similar presentation maybe.
• 02:18I actually don't remember maybe 5 years ago. It's somehow if you're like yesterday. So I'm Steven and reduces my email just a little bit information. Almost center if I'm a professor about statistiques is Melina mention I direct our biostatistics shared resource and I also work directly on skin cancer and also long cancer. We have 2 sparse and have been part of the skins borrow for 6 seven years.
• 02:49I recently joined the long spar so for both sparrs. I derived bio. Staten about if Mattis core so if you work on those 2 type of cancers. You may have all you well. They rank work with me, so this is a point. I want to highlight so if you need any cancer. Analytics support, Val Statistiques by Formatics or anything.
• 03:16Dealing with numbers, which may or may not be classified told me my office in search for the very close to here just about maybe 5 minutes walking distance and so you're Welcome to stop by it. How to me at any point and well. This may be the best way to reach me my email. You only respond within 24 hours, so if I don't.
• 03:42Well, that means there's something horribly wrong, so check mark check with me.
• 03:47OK. So who we are, we are one of the shared resources are actually a bunch of those shared resources actually lost con sometime ago and we sort of land by doctor David Stern. He sent me email saying he's really sorry. He can be here today. Here's kind of center celebrated writer for sharing resources and my shared resource we stationed unit Department.
• 04:18Dial Statistiques and we see in school public house and also we are sort of part of Wack ass stands for yield Center for analytics Sciences, so if you work on cancer directly to me if your research is more related to other type of diseases diabetes. Cardiovascular diseases anything else, you're also Welcome to talk to white casts basically.
• 04:50Find my team we do cancer for anything else that will be like ass.
• 04:55So we have been supporting those large team studies like the sparse nowadays. Other pay grants. But we also work with individual status. Our one standards are twenty one R 03 or pilot studies. So big found a small standards totally awesome will be equally high pay to herb so exciting one we do, we provide?
• 05:22I don't want to say 100% but maybe in 99% of bile. Statistique also part and we also do some by dramatic support. It's more sometimes maybe a little bit inconvenient for investigators since there are several bounding formatics groups on campus, including us, including.
• 05:44As some people Department pathology another body formatics all makes related groups and actually. I don't know how many are there, but maybe a good idea to start with me and if we can help you we will do that. If we cannot. I will direct you to the right person. So not bad idea to start with with us and also we do some data management, especially high end.
• 06:15Instrument if you have bunch of numbers on the paper. You want to put them in Excel sheet. Maybe not a good idea to talk to us. I mean, we can do that. But honestly I don't like doing that. But if you have large complex databases. You want to manipulate to us. I mean, we're still building. Our data management capacity. So we can do some but maybe not all well again start with us.
• 06:46So are clearly introduction of our team my surf. My all expertise is genetic Epidemiology so trying to look at what kind of risk factors genetic non genetic may contribute to cancer and I also do general balance that has to access implies largest regression linear regression and we also have turns out man who is a professor in Bar Statistiques, Here is really great aspart in clinical trial, he also does.
• 07:16PRC protocol reveal so for designing on analyzing clinical trials, he may be the best resource on campus.
• 07:27We have tired time is has been providing populations that is support also is really worldwide. Renew esport in lung cancer. So maybe 2 or 3 years ago, he went to the White House.
• 07:50And it may be longer than than it should be the last president so he went to the White House calls. Some really prestigious award for his work in lung cancer and we also have data way way is here. Here's a new assistant professor also specializing clinical trial. General Bouscat Estates. He also does part of PRC reveal we couple years ago. We proud him. Michael, who works in this very.
• 08:21Then your science very sexy area biomedical bake data so apparently if you have this word in your proposal. The hours of guide founded just chunk hub so if you have any ideas related to big data taught Michael and the daughter. Young is research scientist specializing in Barrie Formatics and we also have punctual master level biostatisticians, who work on a lot of different cancer types.
• 08:51Different types of analysis and I think combined together, they have, maybe 2530 years of experiments so really experienced status station.
• 09:02And just a little later advertisement and we're also trying to go. Look for a pawn in for mentation. So if you have any candidate any of your collaborator priest out once an lisha is part time admin person so she helps us keeping track of all projects and I think this is bigger advantage. We have, we have a huge Department files that has tax so if you need expertise, not covered by us.
• 09:34Play happy you'll build collaborations.
• 09:38OK So what we can do on one we have been doing we can help you design your study so this is called they passed from a proposal. We recently date and we wrote this and we also Dayton is actually quite fancy plot so basically we can tell you where if you start was a summer size. You may need efforts coming will try out, you need any interim analysis, an want the end result will be.
• 10:08Also to commence us caters to commence reviewers, we can do analytics correctly.
• 10:16So you may know the study. We are also there. We want to make sure the study is done correctly data collected biased way, and all procedures are followed so there will be no surprise after data collection if there is any plan for interim data analysis. We will do that and will report results and sometimes the actual results are just different from what you may expect it in this case, we will read.
• 10:48Visit honey revise your data analysis plan if needed. Honest believe or not, and this happen actually quite commonly so you do need very strong very convincing way to tell an age tell earn SQL data. Those are the software packages. You need nodes under steps so this is for the sake of the proposal and also to convince us caters to converse reviewers.
• 11:18We can do analytics correctly.
• 11:22So, in the middle of study. We are also there. We want to make sure the study is done correctly data collected biased way, and all procedures are followed so there will be no surprise after data collection if there is any plan for interim data analysis. We will do that and will report results and sometimes the actual results are just different from what you may expect it in this case, we will.
• 11:53Visit honey revise your data analysis plan if needed. Hornists believe Arnard and this happen actually quite a commonly so you do need very strong very convincing way to tell age tell FDA OK. I need to change my data collection on my data analysis plan and we can do that for you.
• 12:17So we more after data collection will do sorrow data analysis and so far. We have been doing 2 things. The first one, is using existing techniques. Existing software so doing Biostatistics Bayleaf. Matic analysis sort of standard way, and this is what I want to highlight so if you look at our research profile. Our bio style Department and we're actually doing quite while so.
• 12:48Beyond using existing standard methods and a software we can also develop techniques on the software tailored to your static. So, your study will be the 1st to use those cutting edge really novel analytical methods and that actually help your publication hype. Your future proposals. A lot, so here are some examples. One this one is for Omics Data Genetic Genomic.
• 13:20Bing actually you're leading role for those network based analysis. So instead of looking at individual jeans. We now look at generating networks and this is sort of a new item technique. We recently developed in pretty much all kinds of studies. We have those small slides from biopsies and or so many biopsies biopsy samples and so many imaging data just there so we recently started you're looking at.
• 13:52Anna CF weekend add more information for example, for prognosis for treatment selection based on imaging data.
• 14:01One is only job half down, we also help you with manuscript abstract and the posters. So we can write your status. Call analysis section like this we can help you with tables charts figures like those and.
• 14:20Well, I have been in this business long enough, but I I don't think I ever have a paper got accepted without revisions. So this will always happen. You always have nasty comments from reviewers. So we will be there for you address those comments.
• 14:39So also we can help you I mean after your static. I'll founded after paper get published one else we can do. Can we have additional papers from the same data can, we have follow-up grant applications.
• 14:57And I have some old data, while this from fiscal year 2017, so we
• 15:04While this simply won this number is a number of cattle center members we directly work with well, just imagine.
• 15:15Sort of behind each member with directly work with now is that Kim so we interact maybe half or model half. Wise you see members and one this is a too small to see in but in slightly over a year I mean, we contributed.
• 15:361680 I actually don't remember the number, but a big number of Cancer Center Publications. So I'm actually quite a problem. We are everywhere in the Cancer Center.
• 15:49So our policies.
• 15:53Start with me start with Alicia so if you have an idea have any question send an email to me to Alicia and let me know what we need, and again, you really we respond within 24 hours if it's something we can do. We will assign about static stational by orientation, he or she will Skype to a meeting with you guys to know more about your program. It's exactly what you need an also maybe give you a estimated timeline like we need this number.
• 16:25Some kind of stung by that time and if it's beyond our expertise for somebody for Mattix. For some data management. We will tell you who do content so bottom line is you will have a response and there will be follow-up meetings. We will have again. We will have a status Stational Bar information assigned to your project and we will follow up until.
• 16:56Final product can be a poster. A paper found data study, or more often, or we try and head study so but anyway, we will work with you until the end.
• 17:14So.
• 17:16Well, this is our official policy, and this is a written on the website and this is written in our grand renewal submitted to NCI so if you have a good idea for proposal. We will try to contribute as much as as you need and this is apart Atom being this is the part. I really like about NCI. They give us more not a big amount of money, but enough money for us to provide you.
• 17:47Am uncle freeze apart again this is free so for bow statistiques, each of you can get up to 16 hours for value mattix. We're we can provide 8 hours, sometimes more and for service beyond this limit officially we ask for this amount per hour. Let me say 2 things first of all this is much, much lower than outside consultants.
• 18:18Someone from a commercial company.
• 18:21This number is much bigger and also I want to say.
• 18:26In my serve under my team we are researchers so.
• 18:30I mean, we let me say I like money, but has an odd is our first thing on my list. So if you have a big idea but you don't have enough money. Still, tired class. We definitely don't want any single pivot, not published simply becaus of this sort. How do I ask and if you really need a lot of support but again if you have a phone in problem. We do have a facial oversight committee and then they can give us a facial approval for.
• 19:01Asking for money or
• 19:03Offering her pad reduced rate so again. We do have this official policy, but that should definitely should not be our biggest concern.
• 19:17So a couple quick remarks.
• 19:20I don't think any cancer paper can't published simply because of good statistiques. It doesn't happen. But if you will have allowed this statistiques bad analysis plan or?
• 19:36Sometimes just the bag table about a regression model. It's sufficient to sync your papers and proposals. So let's make sure this does not happen and also don't help us last minute. Sometimes I got emails.
• 19:5410:00 PM user night, saying I I have a title on 8:00 AM tomorrow morning and basically the only thing I can say is kholoud so don't do that got us involved early on an we won't be there, so go to my office directly or Welcome to do that and my emails my email. Alicia's email are here and also you see our list of status stations so.
• 20:25If you want to talk to anyone specific do that, so again will be there and I'll be happy to answer any questions, you may have thank you.
• 20:41Any question.
• 20:45Yeah, so I've been fortunate to use the Cancer Center 5. Scott shared resource for 15 plus years and it's it's true. I mean, you can if you get up a paper you know revise and resubmit and you're lucky and then if it's not appropriate statistics. It'll get rejected and so it's really critical to have that by statistics support not only in analysis, but also submitting the grant, and then once carrying out the study so please reach out to them if you need their help.