146 replies to this topic

[Mind]

• Proposal: "Approximate Replication of Baati, et al C60 in Olive Oil
• Team Leaders: S123, Mind, AgeVivo
• Leader workload: 60-120 minutes per week with most of the workload focused in the early stages of the study, screening candidate pet owners, shipping C60 and placebos.
• Team Members: 3 to begin with (leaders), 4 if you include the C60 supplier.
• Members workload: Unknown. We suspect there will be some work involved in making sure experiment participants post regular updates in the Longecity forum. Some of this might be handled by moderators or BP
• Members only? No
• Funding Required? - Yes.
• Funding Level: $500 We are applying for a "small grant" from Longecity. Up to 100 C60 samples will be provided for free. Shipping is probably going to be the main cost. Experiment participants will have to procure their own animals, if they don't have them already.
• Metrics for evaluating success or failure: Not completely fleshed out. As an initial metric of success, I would suggest having a sample size twice the size of Baati make it through the end of the experiment.
• Milestones / Interim Steps: Perhaps S123 or Agevivo can suggest some things here.

Read here for some of the initial discussion of this project.

It has been decided that AgeVivo will produce the placebos and randomize the samples. 10 C60 samples have already been shipped.

We realize we cannot replicate the Baati study exactly with this project, but we hope to get somewhat robust data, while also helping build a community of citizen scientists collaborating through Longecity.

[AgeVivo]

Mind, Well done as always!
In the leaders wworkload we also have "Producing the placebos" of course (I'll play with the colorant suggested by Sven; perhaps I'll need to try another one. It might decay the start of the experiment by a few weeks, which is probably even better not to rush too much once pet owners are ready with their animals).

Your metric is good I think, I have no idea of whether it is ambitious now. I should know in the coming days/weeks.

[AgeVivo]

• Milestones / Interim Steps: Perhaps S123 or Agevivo can suggest some things here.

Here is a proposal:

• 1 prepare the experiment with two parallel tasks: a) and b) in parallel:
  
  • a) receive samples, check them and set up the experiment from a solution point of view
    
    • Find how to make a correct control. Here we need to test the colorant that S123 has proposed on carbon's samples.
      As sent to you by email I know several persons who can be of good advice here
    • See how long it takes to receive the solutions, if it is easy to label them, send them/ other potential things as we test the process.
    • Tests the solutions
      Try to detect potential chemicals, other that comes to mind. Eg weight of drops
    • I imagine this could be done by mid/end of september
  • b) find a list of pionniering pet owners
    
    • Ask in different places if people want to participate in the C60 pet-owner experiment
      Described below
    • Help them get ready
      For pet-owners who start from scratch, I can lead them to buy a cage and all the equipment (what I have works for mice or for rats). I imagine that the grant could serve for that (for the equipment, not the animals; animals are just a few $ and it is a personal responsibility). For all pet-owners, we need to test the system before the treatment starts: they take pictures and videos, put them on LongeCity, we send them a little of other olive oil not-to-used solutions (another type of oil without colorant) and check that they receive it well and that they can put it on bread and that the mice eat it.
    • Decide how to choose who is control versus who is treatment
      Pure randomness is suboptimal in terms of statistical power. Constrained randomness is better: for example we can pair pet-owners by similarities (type of animal and conditions, male/female and age at start; rather than pairs, making a group of 4 is better so that it would be very confusing for me to guess who has what when someone has mice dying) and then you can roll a dice (/ask a program) to randomly choose who among the two have a treatment.
  • Note: unforeseen needs may come at this stage, so this stage is very important for the experiment not to be useless
• 2. Start treatments with pionniers
  
  • The hope is to have sufficiently well prepared the experiment so that everything is easy
  • In science details count and lots of things can happen so we need to be attentive => it is good to start with a few persons
• 3. Gradually increase the size of the experiment
  
  • If it works well with pionneers then the hope and guess is that other people will join
    When proposing MPrize at home many persons told me they would be interested after others start (I hope that writing this here won't transform the pionniers in followers)
  • Keep the pool concept
    Decaying things in time creates a changing environment in terms of seasonality. This can be seen as a bad or a good thing, just like working on one particular strain in very specific conditions, or in the contrary trying to work on a pool. Here, the concept is the pool.

Feel free to discuss it.

Edited by AgeVivo, 14 August 2012 - 08:12 PM.

[smithx]

Decide how to choose who is control versus who is treatment

Pure randomness is suboptimal in terms of statistical power. Constrained randomness is better: for example we can pair pet-owners by similarities (type of animal and conditions, male/female and age at start; rather than pairs, making a group of 4 is better so that it would be very confusing for me to guess who has what when someone has mice dying) and then you can roll a dice (/ask a program) to randomly choose who among the two have a treatment.

It's really important for the experiment to be double blinded if we want it to be taken seriously. This means that neither you nor anyone involved connected with the test subjects knows who got the placebo and who got the active compound.

The key to who got what should be determined by a separate person and kept secure by them and a few other otherwise unconnected people, and should be revealed only after the experiment is over.

You should not even try to guess who has what when someone has mice dying. The data should just be entered and kept until the experiment is entirely over. Only after all the mice have died should the data be analyzed.

This is the only way to avoid any hint of tampering and to keep the study entirely objective.

I also suggest that every person participating be required to complete a bi-weekly online form updating the status of their animals. The status update should include whether or not each animal is still alive, and perhaps a couple of multiple choice questions about the health status of the animals. It should also include a few questions regarding whether there were any changes in feeding or other conditions during the last two week period.

By having regular participation, we should decrease the percentage of people who lose interest and whom we never hear from again.

[AgeVivo]

Thank you Smithx. Absolutely Absolutely Absolutely!

Double blind: I agree. Too often research is not double blind. This WILL be double blind. I am in charge of randomizing the samples... but I am not planning to know who has what! I don't wnat to know, it could lead to some bias! So, I am planning to consider groups of pets under similar conditions, and make a little computer program that randomly decide who among those pets will receive treatments or placebos. To have someone independent put letters on the bottles and tell me which bottle to send to which participant, and he will make the information officially available for the end of the experiment (and that person would be observed by another person, to avoid issues). I have started to think about the details -- if you want to discuss it/propose something specific we can make a specifc thread for that: it is not specific to mprize at home with c60). As a result Mind and I could be participants for example.

Randomization knowns only after the experiment is over: absolutely. Too often research is not double blind.

bi-weekly online form updating the status of their animals: yes but bi-weekly is too frequent and the form will be a post with images at least as often as changing animals: In most cases the age of pets won't be known more precisely than one week, and here we don't care about a precision of one week in lifespans (not interesting and far too small N to detect it anyway). So we are asking to take a picture of each animal during the feeding and to post it along with a comments. We are also asking to take a picture of all animals along with the water bottle, to make sure it was changed as well. To avoid receiving pictures that where taken at another date (at some point it impossible to control everything), at anytime we can ask anyone to take a picture in a specific position (e.g. with a pen on the top of the cage...). If at anytime someone forgets to change the cage, he will be immediately reminded by email and PM (at least).

This will bring much information. For example the arrangement of the hair tells a lot about the health of an animal (an ill or stressed animal has shaggy hair) and we will see the bedding of the cage (if it has been changed, without food nor dung). So it will help me/we advise pet owners precisely. and collectively tune settings at the beginning. I prefer to let them free to write what they truly think in their posts with images than constraining them to a format that may lead them to see things differently than they would otherwise. A little like what is happening with my current experiment.

Edited by AgeVivo, 15 August 2012 - 09:45 PM.

[AgeVivo]

PS: the great thing with open science is that you are tracked like no researcher is his lab. And learn fast. Especially in forums where criticism is very spontaneous. See my current experiment for example, where I was not well aligned with the rat dose of the paper. My feeling is that this can compensate a lot of amateurishness, especially if "open science" concerns the preparation of the experiment (including have every pet owner post images and comments online before the true treatments are started). We will see... and hopefully learn a lot from it.

Edited by AgeVivo, 15 August 2012 - 11:12 PM.

[AgeVivo]

The experiment will lead to continuous discussions and refinements, so while main topics to think about can be mentionned here, I made a thread where they can be discussed in great details:
http://www.longecity...us-refinements/

The randomizing will be done and videotaped by "someone"; we won't know about it until the video will be put online after the very last mouse has died.

Survival statistics: the LogRank test will be performed, as indicated here: http://www.longecity...val-statistics/

Edited by AgeVivo, 16 August 2012 - 07:50 PM.

[niner]

A couple comments:

The key to the randomization should be backed up in several locations so that a disc drive crash or some other disaster doesn't wreck the whole experiment. At least one backup should be non-electronic.

Asking people to post photos is a lot of work, which might make them bail out after a few months because it's too much trouble. I would ask for photographs much less often, like once or twice a year.

Regarding not breaking the blinding until the last rat is dead, I'm not sure we need to go that far. Once the animal is dead, it's dead, and whether or not anyone knows it got the active or the placebo isn't going to change that. The only way information could leak out would be if half the animals were dead and they were all controls, then everyone would know that the remaining animals were active. We could share the information among ourselves without corrupting the experiment. The key is that pet owners with living animals shouldn't know what they have. The pet owners may figure it out eventually, anyway, if their animals live significantly longer than they "should".

[smithx]

I think it is important that one person do the randomization and another send the samples out.

The person who sends the samples should have no way of knowing who got what.

[AgeVivo]

I think it is important that one person do the randomization and another send the samples out.

The person who sends the samples should have no way of knowing who got what.

Yes, absolutely. Perhaps it was not a good idea to create another thread (http://www.longecity...us-refinements/) because I have already answered about it there.

[AgeVivo]

The key to the randomization should be backed up in several locations so that a disc drive crash or some other disaster doesn't wreck the whole experiment. At least one backup should be non-electronic.

Yes. Again I have created a spread specifically do discuss in details: http://www.longecity...us-refinements/

Asking people to post photos is a lot of work, which might make them bail out after a few months because it's too much trouble. I would ask for photographs much less often, like once or twice a year.

Text:
- One post (text) at each cage change is required I think (pictures or not), just for the record of the experiment and to react rapidly if anyone was forgetting the pets
- If there are exceptions (3 week vacations for exemple) it should be clearly explained in advance / discussed with Mind and me at least, to have some securitization of the experiment.

Pictures:
At current numeric times I find that taking a picture and post it online (or to me by email) is not very difficult so here is the minimum requirement I see:
- Pictures before crucial in the few weeks before the start of treatments, to check that everything is all right, to tune things together and be able to compare the "after" treatment with the "before" treatment => once a week, during that time
- Then during the fist month of treatment I think it is important too, for the same reasons => once a week during one month
- Then I agree with you, though once a month seems nice to me
- Once the pets are 18 month old I would recommend more frequent than one month though.
I imagine that on a case by case basis it can be discussed with pet owners and adaptation can be mentionned online. I just think that having a clear record of the experiment is important.

Regarding not breaking the blinding until the last rat is dead, I'm not sure we need to go that far.

If someone was ever cheating and reporting no death during 10 years, we wouldn't like to wait until his mice died ;-) So far the list of participants seems very serious to me, I hope it will continue a such. It might be sufficient indeed to undisclose who had what at the end of one's experiment, but perhaps it is too loose: I have to think about it in the coming say 2-3 weeks, depending on how difficult it will be to distinguish colored oil from C60 in olive oil, and depending on the list of pionneers.

[AgeVivo]

Groups. Based on the preliminary view of a dozen participants that I have listed, with one third of rat owners (males and females from different ages) and two thirds of non-pet owner currently -- the first informed persons here are mostly not pet owners, so now the share of pet owners is growing -- here are the groups that I foresee:

1- pretty sure: replicate Baati et al in female rats (Baati et al was in males)
2- probable: replicate Baati et al in male rats (probably smaller N than Baati et al)
3- probable: a couple of mice, males or females, a little like what I currently do

The age at start would be the same for all, and slightly older than 10 months as done in Baati et al. All this to be confirmed depending on who participates; but I thought it would be good for this "open science" (, double blind and multi centric) experiment that share this preliminary vision with you.

[AgeVivo]

Treatment or placebo: actually *Perhaps* both in some pet owner homes; but with some constrains to reduce risks.


1) to largely cancel variability across lifespans across homes

I have longly discussed with a former researcher colleague who has a lifelong experience on animal studies (btw he proposed to test this week which colorants can be used to make the placebo). In long term experiments that are not in specific pathogen free conditions, he has seen -- and there are such reported cases even by the greatest mouse specialists -- that sometimes animals have small, long lasting disease: one would not perceive it just by looking at the animals, but when you study a specific organ (liver, lungs...) you sometimes discover that something is strange in your group of animals: you investigate: your animals have had a disease. Certainly we ourselves are full of diseases (and full of diseases that can be transmitted from humans to humans). This should creates some variability in the lifespans, especially across homes.

For that reason (and also because of difference in care and other differences in the animals or environments) he believes that the statistical power would be improved if we can compare treatments and placebos within a home, even if one or two persons mix treatments. Of course, this is a trade off, and of course if c60 really increases lifespan by very much, who cares.


2) how to reduce risks of giving treatment instead of placebo

a) animals should be clearly distinguished between the two cages
Using a marker like ear whole or tatoing... is difficult and not adapted for amateurs. But one can use color caracteristics: the animals should be placed in two different cages; the two cages should have different colors, and for each cage pictures should be able to highlight very clearly the color caracteristics of the animals. If two animals have similar color caracteristics they should be placed in the same cage. The researchers believes that if you put mostly white rats in cage1 and most black rats in cage2 (and that it is not a pure black strain in cage1 and a prue white strain in cage2), then the a priori differences in lifespan should be zero (zero for real or zero compared to the not very big number of participants for example)

b) "treatment of placebo"1 (resp 2) should be in a bottle with the same color as cage1 (resp 2). Same for instruments.

And it should be made very clear that one should never use instrument1 for cage2 for exemple. That way, it becomes really difficult to make a mistake by inadvertance. And because the participants do not know what is treatment and what is placebo, and because both treatments would a priori (...we are checking) extend lifespan, to different degrees, there would be little incentive to mix (and if ever the results are clear: some animals living >3 years, cheats would be easily identified and known)

c) What's more, from the email exchanges that I have had so far, I have had the great surprise that all the persons interested in participating look serious.


In practice
I preferred to inform you about this possibility, so that you can react (and say: no way, or yes, or why not, or let's see later...) but it will depend on the participants:
- how many cages they have already (and whether they want to / can have more; in particular a matter of available room and time to clean cages)
- if the rats of pet owners can be clearly identified and separately as desired for the experiment for two cages (it might be better for the animals to keep them together)

So, to be seen...

[Mind]

sorry about the new thread, but a specific proposal for a $500 grant had to be posted for the Board's approval. We can merge these discussions once we have the approval.

One thing I wanted to make sure everyone is on the same page about, is that a prospective pet owner will only receive placebo OR C60. Some of the people who have contacted us about this are very enthusiastic and would like to have several rats at home, giving some placebo and some C60. I have stated that they will only get one OR the other. That way we reduce the possibility of user error. If people want to start their own small scale experiment at home and post about it here, that is fine, but for our Longecity-sponsored research, they must accept that they will get only placebo OR only C60.

Edited by Mind, 19 August 2012 - 11:32 AM.

[Logic]

Treatment or placebo: actually *Perhaps* both in some pet owner homes; but with some constrains to reduce risks.


1) to largely cancel variability across lifespans across homes

I have longly discussed with a former researcher colleague who has a lifelong experience on animal studies (btw he proposed to test this week which colorants can be used to make the placebo). In long term experiments that are not in specific pathogen free conditions, he has seen -- and there are such reported cases even by the greatest mouse specialists -- that sometimes animals have small, long lasting disease: one would not perceive it just by looking at the animals, but when you study a specific organ (liver, lungs...) you sometimes discover that something is strange in your group of animals: you investigate: your animals have had a disease. Certainly we ourselves are full of diseases (and full of diseases that can be transmitted from humans to humans). This should creates some variability in the lifespans, especially across homes.

For that reason (and also because of difference in care and other differences in the animals or environments) he believes that the statistical power would be improved if we can compare treatments and placebos within a home, even if one or two persons mix treatments. Of course, this is a trade off, and of course if c60 really increases lifespan by very much, who cares.


2) how to reduce risks of giving treatment instead of placebo

a) animals should be clearly distinguished between the two cages
Using a marker like ear whole or tatoing... is difficult and not adapted for amateurs. But one can use color caracteristics: the animals should be placed in two different cages; the two cages should have different colors, and for each cage pictures should be able to highlight very clearly the color caracteristics of the animals. If two animals have similar color caracteristics they should be placed in the same cage. The researchers believes that if you put mostly white rats in cage1 and most black rats in cage2 (and that it is not a pure black strain in cage1 and a prue white strain in cage2), then the a priori differences in lifespan should be zero (zero for real or zero compared to the not very big number of participants for example)

b) "treatment of placebo"1 (resp 2) should be in a bottle with the same color as cage1 (resp 2). Same for instruments.

And it should be made very clear that one should never use instrument1 for cage2 for exemple. That way, it becomes really difficult to make a mistake by inadvertance. And because the participants do not know what is treatment and what is placebo, and because both treatments would a priori (...we are checking) extend lifespan, to different degrees, there would be little incentive to mix (and if ever the results are clear: some animals living >3 years, cheats would be easily identified and known)

c) What's more, from the email exchanges that I have had so far, I have had the great surprise that all the persons interested in participating look serious.


In practice
I preferred to inform you about this possibility, so that you can react (and say: no way, or yes, or why not, or let's see later...) but it will depend on the participants:
- how many cages they have already (and whether they want to / can have more; in particular a matter of available room and time to clean cages)
- if the rats of pet owners can be clearly identified and separately as desired for the experiment for two cages (it might be better for the animals to keep them together)

So, to be seen...

+1

I think its very important for both the rats/mice recieving C60oo and Placebo to have the same husbandry, enviroment, food etc. and to get the same infections etc if any.
Else you may as well give everyone the real thing as the average lifespan of rats, mice, etc. is already known.

How difficult can it be to keep dark mice in a dark cage (paint) and feed em meds from a bottle with a dark lable without getting mixed up with the light/white rats, cage, meds?
If labs can do it (protocol); surely we can?

[Mind]

+1

I think its very important for both the rats/mice recieving C60oo and Placebo to have the same husbandry, enviroment, food etc. and to get the same infections etc if any.
Else you may as well give everyone the real thing as the average lifespan of rats, mice, etc. is already known.

How difficult can it be to keep dark mice in a dark cage (paint) and feed em meds from a bottle with a dark lable without getting mixed up with the light/white rats, cage, meds?
If labs can do it (protocol); surely we can?

I understand your reasoning, but we are doing something new here, outside of a lab environment, and there will be much more opportunity for error. Just think about all the things that go on in your house. Pets, kids, spills, late nights, etc... I would rather not introduce more opportunities for error and make sure all of our sponsored researchers only get either C60 OR placebo. My thought is that with a large sample size, the differences in husbandry will be evened out. We will give people care tips and a feeding protocol to even things out as well. Remember that we are trading off some of the "strictness/protocol" of a lab environment in order to build a more solid community of people advocating healthy life extension.

If everyone else thinks we can give individual pet owners both placebo and C60 with no worries, then that is the way we will do it. I am just letting you know my reasoning for being against that method.

[smithx]

One thing I wanted to make sure everyone is on the same page about, is that a prospective pet owner will only receive placebo OR C60. Some of the people who have contacted us about this are very enthusiastic and would like to have several rats at home, giving some placebo and some C60. I have stated that they will only get one OR the other. That way we reduce the possibility of user error. If people want to start their own small scale experiment at home and post about it here, that is fine, but for our Longecity-sponsored research, they must accept that they will get only placebo OR only C60.

I strongly agree.

Any other approach is asking for trouble and will make any results we get much more highly questioned.

Ideally all animals would be treated identically, but when randomized across a sufficiently large sample size, any difference in the treatment of any particular set of animals in one household will become insignificant.

[smithx]

I'd like to add that if Agevivo is dealing with the pet owners, someone else should prepare the all samples (both C60 and placebo) and randomize and label them before sending them to him.

Whomever is sending samples to pet owners should get a box of samples that all look alike, with sequential numbers on the bottles (1, 2, 3, 4). Sample 1 may be placebo, or maybe 2 is, or maybe 1 and 3... no one should know this except the person who prepared them and who keeps the key.

This is what it means to be double-blinded. The person conducting the study should have no way at all of knowing which samples are C60 and which are placebo.

The key should not be disclosed to anyone until the end of the test period, and the test period should not end until most of the animals are dead.

If any of these measures are not taken, the study will not be a reliable study.

[AgeVivo]

Hi Smithx,

Thank you for trying to make this experiment as great as possible. You are obviously sensitive to important points in the design of an experiment and this is great. I am trying to combine all this with the limits of extremes: here is some detailed information so that you can help me define the design with detailed elements in hands:

[Mind]

Thanks for the comments SmithX and Logic. One way to balance the XOR & AND of substances sent to the community experimenters would be to evaluate some of the people who would like to conduct a small trial of the placebo & C60. Perhaps we can strongly suggest to most of the experimenters that XOR (either placebo or C60) is the way to go and it is through this method that they will make the greatest contribution to longevity science. For very enthusiastic experimenters who have a good set-up at home, and feel strongly they can maintain segregation of the animals, we could allow it - but then we would emphasize that they should report/log any possible anomalies.

[Logic]

Thanks for the comments SmithX and Logic. One way to balance the XOR & AND of substances sent to the community experimenters would be to evaluate some of the people who would like to conduct a small trial of the placebo & C60. Perhaps we can strongly suggest to most of the experimenters that XOR (either placebo or C60) is the way to go and it is through this method that they will make the greatest contribution to longevity science. For very enthusiastic experimenters who have a good set-up at home, and feel strongly they can maintain segregation of the animals, we could allow it - but then we would emphasize that they should report/log any possible anomalies.

Thx Mind
That would be great, but I dont want to upset the apple-cart. So whatever Longecity decides will be fine.
Has anyone else shown an interest in having both C60 animals and controls?

[Mind]

We appreciate your input very much. Keep it up.

Has anyone else shown an interest in having both C60 animals and controls?

There is only one that I am aware of so far.

[AgeVivo]

DOUBLE-BLINDNESS
As I said: "This WILL be double blind"

I'd like to add that if Agevivo is dealing with the pet owners, someone else should prepare the all samples (both C60 and placebo) and randomize and label them before sending them to him.

I know a researcher who uses many colorants. He nicely proposed to test which one(s) to use in olive oil. He told me it is not easy:

• the color is not easy to reproduce with usual colorants
• one way would be to use colorants on both solutions but it is difficult because fullerenes have many interactions with colorants
• many colorants have biological effects (at least because they can be hypophobic or hydrophilic for example) so small doses are needed

I don't want to know more details, in order not to have clues to recognise what is what. I imagine the process then as follows:

• I should receive the treatments and empty bottles soon
• I will wash the empty botlles very well to ensure they are clean, and will not look at the inside ot the treatments
• I will bring the bottles (empty and not) and the olive oil to the lab of my researcher friend
• Another person of trust will be there and will videotape things and movements as he desires
  
  • I am now going to PM you some details in case you have a person of trust in this area
  • I don't want to know the name or address of that person of trust
• The researcher will prepare the samples and randomize them
  
  • He will put letters "ABCD" or "ADCB" etc. instead of "placebo placebo treatment treament". I will have defined the "ABCD" "EFG"... groups in advance, and this will be posted online
  • He will put the randomisation on paper twice. Once for him, once for the other person: both will keep it secret until the end of the experiment
• I will get the bottles, with their letters on it. I won't look at the inside
• I will send them to the pet owners. To allow confidentiality if desired, only Mind and I have the addresses of the pet owners.

If you wish, please propose something better. Main constraint: it would be terrible if for some reason it is not known at the end of the experiment who has received what

Edited by AgeVivo, 20 August 2012 - 07:51 PM.

[AgeVivo]

Thanks for the comments SmithX and Logic. One way to balance the XOR & AND of substances sent to the community experimenters would be to evaluate some of the people who would like to conduct a small trial of the placebo & C60. Perhaps we can strongly suggest to most of the experimenters that XOR (either placebo or C60) is the way to go and it is through this method that they will make the greatest contribution to longevity science. For very enthusiastic experimenters who have a good set-up at home, and feel strongly they can maintain segregation of the animals, we could allow it - but then we would emphasize that they should report/log any possible anomalies.

Thx Mind
That would be great, but I dont want to upset the apple-cart. So whatever Longecity decides will be fine.
Has anyone else shown an interest in having both C60 animals and controls?

There is only one that I am aware of so far.

Actually there are 3 of them, who since the start indicated that they would like to compare c60oo and colored olive oil (and versus nothing, at least for 2 of them). Mind, we will exchange with them to see what they can put in place to minimise risks. But first, we need to wait to know if the placebo looks like the treatment (without that we look at the solutions ourselves, of course).

I would like to underline that it would actually be great to have placebo AND treatment in some well set-up homes:
1) the number of pet owners will be limited (perhaps it is "indefinite" at these early time, but certainly not "infinite"!)
2) if some home ever has an issue, and that animals die rather young, and that we count it as normal, there is a risk that we get very biased results In that case, the places where there will be both placebo AND treatment will have the most robust signals.

So having both "AND" and "EITHER" treatment/placebo in different homes is a good way not to put all our eggs in the same basket.

Edited by AgeVivo, 20 August 2012 - 08:22 PM.

[smithx]

I really believe that the way you are proposing to do things is less than optimal. If you have only two sets of samples, it will be very likely that at some point someone will figure out which is which, and this again can cause the entire experiment to be questioned. The samples should be COMPLETELY RANDOMIZED, not put into two categories.

The key to which is which should be a spreadsheet, encrypted and then given to several people for safekeeping. At the end of the experiment, the encryption key should be published AFTER all the data is tabulated and published. This again will prove that nothing was tampered with.

From my other post:

- Person A supplies a set of C60 samples
- Person B supplies a set of placebo samples
- Person C randomizes the numbers and puts the randomized numbers on each sample. So sample 1 may be either C60 or placebo, sample 2 may be either C60 or placebo, etc. Only the key lets us know which is which.
- The key should be put into an encrypted file and distributed to several people to insure that it isn't tampered with. The password to the file should be kept secure in one place.

- After this process is complete, Person D gets a box of samples which are completely randomized. This person has no way of knowing which is which.
- Person D sends the samples to the test subjects, monitors them and collates the results

- The experiment should have a defined time limit longer than the expected lifespan of any of the mice, or end when all the mice are dead, whichever comes first
- After the end of the experiment, the data should be analyzed and published
- Only after this step should the key be revealed, so that we can see if the lifespan of the mice was correlated in any way with whether they got placebo or C60.
- The fact that the data is analyzed before knowing the key, and the fact that they key was held by several people and therefore hasn't been tampered with, will prove the validity of the results.

[AgeVivo]

I really believe that the way you are proposing to do things is less than optimal (...) From my other post

... thank you for this 'compliment'.. Sorry I didn't know about the other post, I guess it has skipped to my attention. which post?

Concerning the proposal:

1. Thank you for the encrypting file trick: you would in addition provide the crypted data (video and correspondance between rats and treatments) to numerous persons: good idea. I know that it is easy to find ways to decrypt files with online tools. I think I have read once that the encryption contained in zip file is secure?

2. For the persons, is what I proposed above Ok then?:
Person A = the supplement seller who accepted to send C60
Person B = the researcher
Person C = the additional trust person
Person D = me
It is true that I also receive the C60 from person A, and I also receive the olive oil. But this is in special packages, if I were to open them (which I won't) then person B would notice it. I don't think it is particularly bad but let me know.

3. Could you elaborate on the following? I don't think I have only two sets of samples, unless you describe "C60oo" and "oo", so I guess I don't understand:

If you have only two sets of samples, it will be very likely that at some point someone will figure out which is which"

4. Some categories should be done. I think the discordance comes from the interpretation of what "category" is.

The samples should be COMPLETELY RANDOMIZED, not put into two categories

- Well, I was essentially thinking about the following categories: "male rats", "female rats", "male mice", "female mice": The number of pet owners is not infinite so I wouln't want to have mostly placebos for one of such categories and mostly treatments for another one: comparison wouldn't be ideal...
- Then, I was also thinking that there might be subcategories (but not of 2 elements only, I think that is what you mean), depending on very big specificities that we have not thought of yet: to be seen if such categories seem obviously important once we have a better view of pet owners possibilities.

Edited by AgeVivo, 20 August 2012 - 09:04 PM.

[Mind]

SmithX, the method you outline here seems to be the most robust, but pragmatically, we might have trouble. We might not have enough people in close geographic proximity to get the randomization and mailing of samples done in a reasonable amount of time.

Here is the proposal we have come up with so far, with the people we have available:

1. AgeVivo receives the the C60 samples, olive oil and empty bottles (identical to the already filled C60 samples). He inspects them to make sure they are clean and ready to use.

2. AgeVivo delivers the bottles and C60 samples to his scientist acquaintance (Person B). This person colors the olive oil and fills the placebo bottles, preferably under video supervision.

3. Person B labels all of the bottles with random numbers, creates the master spreadsheet (encrypts it, shares the key with at least two other Longecity leaders, and has it backed-up) and creates a randomization sequence (RS) for sending out samples to the experimenters (as you outlined here and especially here). While it would be nice to send out all of the samples at once or in a very short period of time, it is likely that there will be a few weeks between the first shipment and the last shipment, but that should not be a limiting factor.

4. When we have an experimenter ready with their mice, person B consults the RS and sends either placebo or C60 to AgeVivo. AgeVivo does not know which one it is, only where the sample is going. Person B knows what the sample is but does not know who it is going to (Is this blind enough?)

5. The experiment ends when ALL the mice have de-animated.

6. Data is analyzed and tabulated.

7. The master spreadsheet is revealed.

Some other things I don't want to gloss over.

Each experimenter will be given the feeding protocol developed by AgeVivo.
Each experimenter will be given general tips for proper care of their animals.
Each experimenter will sign a liability release form with Longecity (Longecity cannot be held liable for dead pets or any other negative aspect of this experiment)

Edited by Mind, 20 August 2012 - 09:03 PM.

[smithx]

I hope that "less than optimal" wasn't too harsh.

With regard to complete randomization, my worry was that it seemed as if there were going to be two boxes of samples. One box would be labeled Sample A and the other Sample B, or some such naming. No one would initially know which was the C60 and which was the placebo. The problem with that is that over time, it would become possible to figure out which is which, if there was a big difference in the response seen to either sample. At the least, there would be a temptation to guess.

This is why I am recommending that ALL the samples be labeled at once by the person doing the randomization. So if we have 50 bottles of each, we would end up with one box of 100 bottles, labeled from 001 to 100, and there would be no way to know, without looking at the key, which bottle was which.

Whenever a new experimental subject came on line, they would simply be mailed the next bottle in sequence.

This assumes that:
A) we know roughly how many subjects we will have
B) we limit the study to N subjects where N is the number of bottles we make

The advantage is that this is a true double-blind study with no chance of tampering. This also means that all the samples can be sent to AgeVivo as soon as they are created, and reduces the coordination required as new subjects come on line.

As for zip encryption, it can be broken but is probably good enough. Better would be GNU Privacy Guard or TrueCrypt. See: http://lifehacker.co...ncryption-tools

[AgeVivo]

Ok thank you. It seems I had kept some posts for when I'd have more time, and then forgot to read them: nice:
http://www.longecity...687#entry518687
http://www.longecity...700#entry518700
The researcher told me it is difficult to find a colorant of the correct color thas dissolves in olive oil. He continues to try. If anyone wants to try (with some olive oil; no need to have c60oo for that) or thinks he has the solution, please indicate so (but of course don't tell me what precisely! don't put online what it is!)

Edited by AgeVivo, 20 August 2012 - 10:41 PM.

[smithx]

By the way, please read the original thread dealing with replicating the experiment, as there is new info and input there:
http://www.longecity...live-oil-study/